2017-04-12 16:55:22 +00:00
--[[--
This module detects gestures .
2012-11-11 06:00:52 +00:00
2013-03-26 12:59:47 +00:00
Current detectable gestures :
2017-04-12 16:55:22 +00:00
2022-09-04 00:38:27 +00:00
* ` touch ` ( emitted once on first contact down )
2017-04-12 16:55:22 +00:00
* ` tap ` ( touch action detected as single tap )
* ` pan `
* ` hold `
* ` swipe `
* ` pinch `
* ` spread `
* ` rotate `
2022-09-04 00:38:27 +00:00
* ` hold_pan ` ( will emit ` hold_release ` on contact lift , unlike its two - finger variant )
2017-04-12 16:55:22 +00:00
* ` double_tap `
* ` inward_pan `
* ` outward_pan `
* ` pan_release `
* ` hold_release `
2022-09-04 00:38:27 +00:00
* ` two_finger_hold `
* ` two_finger_hold_release `
2017-04-12 16:55:22 +00:00
* ` two_finger_tap `
* ` two_finger_pan `
2022-09-04 00:38:27 +00:00
* ` two_finger_hold_pan `
2017-04-12 16:55:22 +00:00
* ` two_finger_swipe `
* ` two_finger_pan_release `
2022-09-04 00:38:27 +00:00
* ` two_finger_hold_pan_release `
2012-12-30 05:44:55 +00:00
2013-02-02 06:36:29 +00:00
You change the state machine by feeding it touch events , i.e . calling
2017-04-12 16:55:22 +00:00
@ { GestureDetector : feedEvent | GestureDetector : feedEvent ( tev ) } .
2012-11-11 06:00:52 +00:00
2013-02-26 04:32:46 +00:00
a touch event should have following format :
2013-02-02 06:36:29 +00:00
2017-04-12 16:55:22 +00:00
tev = {
slot = 1 ,
id = 46 ,
x = 0 ,
y = 1 ,
2022-05-05 19:00:22 +00:00
timev = time.s ( 123.23 ) ,
2017-04-12 16:55:22 +00:00
}
2021-04-14 23:00:28 +00:00
Don ' t confuse `tev` with raw evs from kernel, `tev` is built according to ev.
2013-02-02 06:36:29 +00:00
2017-04-12 16:55:22 +00:00
@ { GestureDetector : feedEvent | GestureDetector : feedEvent ( tev ) } will return a
detection result when you feed a touch release event to it .
2012-11-11 06:00:52 +00:00
--]]
2017-04-12 16:55:22 +00:00
local Geom = require ( " ui/geometry " )
local logger = require ( " logger " )
2022-05-05 19:00:22 +00:00
local time = require ( " ui/time " )
2019-03-04 18:01:01 +00:00
local util = require ( " util " )
2017-04-12 16:55:22 +00:00
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- We're going to need some clockid_t constants
local ffi = require ( " ffi " )
local C = ffi.C
require ( " ffi/posix_h " )
2022-05-05 19:00:22 +00:00
-- default values (time parameters are in milliseconds (ms))
local TAP_INTERVAL_MS = 0
local DOUBLE_TAP_INTERVAL_MS = 300
local TWO_FINGER_TAP_DURATION_MS = 300
local HOLD_INTERVAL_MS = 500
local SWIPE_INTERVAL_MS = 900
2019-07-24 12:31:20 +00:00
Clarify our OOP semantics across the codebase (#9586)
Basically:
* Use `extend` for class definitions
* Use `new` for object instantiations
That includes some minor code cleanups along the way:
* Updated `Widget`'s docs to make the semantics clearer.
* Removed `should_restrict_JIT` (it's been dead code since https://github.com/koreader/android-luajit-launcher/pull/283)
* Minor refactoring of LuaSettings/LuaData/LuaDefaults/DocSettings to behave (mostly, they are instantiated via `open` instead of `new`) like everything else and handle inheritance properly (i.e., DocSettings is now a proper LuaSettings subclass).
* Default to `WidgetContainer` instead of `InputContainer` for stuff that doesn't actually setup key/gesture events.
* Ditto for explicit `*Listener` only classes, make sure they're based on `EventListener` instead of something uselessly fancier.
* Unless absolutely necessary, do not store references in class objects, ever; only values. Instead, always store references in instances, to avoid both sneaky inheritance issues, and sneaky GC pinning of stale references.
* ReaderUI: Fix one such issue with its `active_widgets` array, with critical implications, as it essentially pinned *all* of ReaderUI's modules, including their reference to the `Document` instance (i.e., that was a big-ass leak).
* Terminal: Make sure the shell is killed on plugin teardown.
* InputText: Fix Home/End/Del physical keys to behave sensibly.
* InputContainer/WidgetContainer: If necessary, compute self.dimen at paintTo time (previously, only InputContainers did, which might have had something to do with random widgets unconcerned about input using it as a baseclass instead of WidgetContainer...).
* OverlapGroup: Compute self.dimen at *init* time, because for some reason it needs to do that, but do it directly in OverlapGroup instead of going through a weird WidgetContainer method that it was the sole user of.
* ReaderCropping: Under no circumstances should a Document instance member (here, self.bbox) risk being `nil`ed!
* Kobo: Minor code cleanups.
2022-10-06 00:14:48 +00:00
-- This is used as a singleton by Input (itself used as a singleton).
2013-10-18 20:38:07 +00:00
local GestureDetector = {
2014-03-13 13:52:43 +00:00
-- must be initialized with the Input singleton class
input = nil ,
2020-10-18 13:57:40 +00:00
-- default values (accessed for display by plugins/gestures.koplugin)
2022-05-05 19:00:22 +00:00
TAP_INTERVAL_MS = TAP_INTERVAL_MS ,
DOUBLE_TAP_INTERVAL_MS = DOUBLE_TAP_INTERVAL_MS ,
TWO_FINGER_TAP_DURATION_MS = TWO_FINGER_TAP_DURATION_MS ,
HOLD_INTERVAL_MS = HOLD_INTERVAL_MS ,
SWIPE_INTERVAL_MS = SWIPE_INTERVAL_MS ,
2014-03-13 13:52:43 +00:00
-- pinch/spread direction table
Clarify our OOP semantics across the codebase (#9586)
Basically:
* Use `extend` for class definitions
* Use `new` for object instantiations
That includes some minor code cleanups along the way:
* Updated `Widget`'s docs to make the semantics clearer.
* Removed `should_restrict_JIT` (it's been dead code since https://github.com/koreader/android-luajit-launcher/pull/283)
* Minor refactoring of LuaSettings/LuaData/LuaDefaults/DocSettings to behave (mostly, they are instantiated via `open` instead of `new`) like everything else and handle inheritance properly (i.e., DocSettings is now a proper LuaSettings subclass).
* Default to `WidgetContainer` instead of `InputContainer` for stuff that doesn't actually setup key/gesture events.
* Ditto for explicit `*Listener` only classes, make sure they're based on `EventListener` instead of something uselessly fancier.
* Unless absolutely necessary, do not store references in class objects, ever; only values. Instead, always store references in instances, to avoid both sneaky inheritance issues, and sneaky GC pinning of stale references.
* ReaderUI: Fix one such issue with its `active_widgets` array, with critical implications, as it essentially pinned *all* of ReaderUI's modules, including their reference to the `Document` instance (i.e., that was a big-ass leak).
* Terminal: Make sure the shell is killed on plugin teardown.
* InputText: Fix Home/End/Del physical keys to behave sensibly.
* InputContainer/WidgetContainer: If necessary, compute self.dimen at paintTo time (previously, only InputContainers did, which might have had something to do with random widgets unconcerned about input using it as a baseclass instead of WidgetContainer...).
* OverlapGroup: Compute self.dimen at *init* time, because for some reason it needs to do that, but do it directly in OverlapGroup instead of going through a weird WidgetContainer method that it was the sole user of.
* ReaderCropping: Under no circumstances should a Document instance member (here, self.bbox) risk being `nil`ed!
* Kobo: Minor code cleanups.
2022-10-06 00:14:48 +00:00
DIRECTION_TABLE = { -- const
2014-03-13 13:52:43 +00:00
east = " horizontal " ,
west = " horizontal " ,
north = " vertical " ,
south = " vertical " ,
northeast = " diagonal " ,
northwest = " diagonal " ,
southeast = " diagonal " ,
southwest = " diagonal " ,
} ,
2022-09-04 00:38:27 +00:00
-- Hash of our currently active contacts
active_contacts = { } ,
contact_count = 0 ,
-- Used for double tap and bounce detection (this is outside a Contact object because it requires minimal persistance).
previous_tap = { } ,
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- for timestamp clocksource detection
clock_id = nil ,
2022-05-05 19:00:22 +00:00
-- current values
ges_tap_interval = time.ms ( G_reader_settings : readSetting ( " ges_tap_interval_ms " ) or TAP_INTERVAL_MS ) ,
ges_double_tap_interval = time.ms ( G_reader_settings : readSetting ( " ges_double_tap_interval_ms " )
or DOUBLE_TAP_INTERVAL_MS ) ,
ges_two_finger_tap_duration = time.ms ( G_reader_settings : readSetting ( " ges_two_finger_tap_duration_ms " )
or TWO_FINGER_TAP_DURATION_MS ) ,
ges_hold_interval = time.ms ( G_reader_settings : readSetting ( " ges_hold_interval_ms " ) or HOLD_INTERVAL_MS ) ,
ges_swipe_interval = time.ms ( G_reader_settings : readSetting ( " ges_swipe_interval_ms " ) or SWIPE_INTERVAL_MS ) ,
2012-11-13 06:12:01 +00:00
}
2014-10-30 18:42:18 +00:00
function GestureDetector : new ( o )
2016-02-16 07:10:07 +00:00
o = o or { }
2014-10-30 18:42:18 +00:00
setmetatable ( o , self )
self.__index = self
if o.init then o : init ( ) end
return o
end
function GestureDetector : init ( )
-- distance parameters
2022-09-04 00:38:27 +00:00
self.TWO_FINGER_TAP_REGION = self.screen : scaleByDPI ( 20 )
self.DOUBLE_TAP_DISTANCE = self.screen : scaleByDPI ( 50 )
2020-10-18 13:57:40 +00:00
self.SINGLE_TAP_BOUNCE_DISTANCE = self.DOUBLE_TAP_DISTANCE
2022-09-04 00:38:27 +00:00
self.PAN_THRESHOLD = self.screen : scaleByDPI ( 35 )
2019-03-05 08:51:00 +00:00
self.MULTISWIPE_THRESHOLD = self.DOUBLE_TAP_DISTANCE
2014-10-30 18:42:18 +00:00
end
2022-09-04 00:38:27 +00:00
-- Contact object, it'll keep track of everything we need for a single contact across its lifetime
-- i.e., from this contact's down to up (or its *effective* up for double-taps, e.g., when the tap or double_tap is emitted).
-- We'll identify contacts by their slot numbers, and store 'em in GestureDetector's active_contacts table (hash).
local Contact = { } -- Class object is empty, as we do *NOT* want inheritance outside of methods.
function Contact : new ( o )
setmetatable ( o , self )
self.__index = self
return o
end
function GestureDetector : newContact ( slot )
-- Check if this new contact might be part of a two finger gesture,
-- by checking if the current slot is one of the two main slots, and the other is active.
local buddy_slot = slot == self.input . main_finger_slot and self.input . main_finger_slot + 1 or
slot == self.input . main_finger_slot + 1 and self.input . main_finger_slot
local buddy_contact = buddy_slot and self : getContact ( buddy_slot )
self.active_contacts [ slot ] = Contact : new {
state = Contact.initialState , -- Current state function
slot = slot , -- Current ABS_MT_SLOT value (also its key in the active_contacts hash)
id = - 1 , -- Current ABS_MT_TRACKING_ID value
initial_tev = nil , -- Copy of the input event table at first contact (i.e., at contact down)
current_tev = nil , -- Pointer to the current input event table, ref is *stable*, c.f., NOTE in feedEvent below
down = false , -- Contact is down (as opposed to up, i.e., lifted). Only really happens for double-tap handling, in every other case the Contact object is destroyed on lift.
pending_double_tap_timer = false , -- Contact is pending a double_tap timer
pending_hold_timer = false , -- Contact is pending a hold timer
mt_gesture = nil , -- Contact is part of a MT gesture (string, gesture name)
mt_immobile = true , -- Contact is part of a MT gesture, and hasn't moved (i.e., would be in holdState if it weren't in voidState)
multiswipe_directions = { } , -- Accumulated multiswipe chain for this contact
multiswipe_type = nil , -- Current multiswipe type for this contact
buddy_contact = buddy_contact , -- Ref to the paired contact in a MT gesture (if any)
ges_dec = self , -- Ref to the current GestureDetector instance
}
self.contact_count = self.contact_count + 1
--logger.dbg("New contact for slot", slot, "#contacts =", self.contact_count)
-- If we have a buddy contact, point its own buddy ref to us
if buddy_contact then
buddy_contact.buddy_contact = self.active_contacts [ slot ]
end
return self.active_contacts [ slot ]
end
function GestureDetector : getContact ( slot )
return self.active_contacts [ slot ]
end
function GestureDetector : dropContact ( contact )
local slot = contact.slot
-- Guard against double drops
if not self.active_contacts [ slot ] then
logger.warn ( " Contact for slot " , slot , " has already been dropped! #contacts = " , self.contact_count )
return
end
-- Also clear any pending callbacks on that slot.
if contact.pending_double_tap_timer then
self.input : clearTimeout ( slot , " double_tap " )
contact.pending_double_tap_timer = nil
end
if contact.pending_hold_timer then
self.input : clearTimeout ( slot , " hold " )
contact.pending_hold_timer = nil
end
-- If we have a buddy contact, drop its buddy ref to us
if contact.buddy_contact then
contact.buddy_contact . buddy_contact = nil
end
self.active_contacts [ slot ] = nil
self.contact_count = self.contact_count - 1
--logger.dbg("Dropped contact for slot", slot, "#contacts =", self.contact_count)
end
function GestureDetector : dropContacts ( )
for _ , contact in pairs ( self.active_contacts ) do
self : dropContact ( contact )
end
end
2017-04-12 16:55:22 +00:00
--[[--
Feeds touch events to state machine .
2022-08-13 22:49:30 +00:00
2022-08-14 23:13:36 +00:00
Note that , in a single input frame , if the same slot gets multiple events , only the last one is kept .
Every slot in the input frame is consumed , and that in FIFO order ( slot order based on appearance in the frame ) .
2017-04-12 16:55:22 +00:00
--]]
2013-03-06 13:04:33 +00:00
function GestureDetector : feedEvent ( tevs )
2022-08-14 23:13:36 +00:00
local gestures = { }
for _ , tev in ipairs ( tevs ) do
local slot = tev.slot
2022-09-04 00:38:27 +00:00
local contact = self : getContact ( slot )
if not contact then
contact = self : newContact ( slot )
2022-08-14 23:13:36 +00:00
-- NOTE: tev is actually a simple reference to Input's self.ev_slots[slot],
2022-09-04 00:38:27 +00:00
-- which means a Contact's current_tev doesn't actually point to the *previous*
2022-08-14 23:13:36 +00:00
-- input frame for a given slot, but always points to the *current* input frame for that slot!
2022-09-04 00:38:27 +00:00
-- Meaning the tev we feed the state function *always* matches that Contact's current_tev.
-- Compare to initial_tev below, which does create a copy...
-- This is what allows us to only do this once on contact creation ;).
contact.current_tev = tev
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
local ges = contact.state ( contact )
2022-08-14 23:13:36 +00:00
if ges then
table.insert ( gestures , ges )
end
end
return gestures
2012-11-11 06:00:52 +00:00
end
2022-09-04 00:38:27 +00:00
local function deepCopyEv ( tev )
2014-03-13 13:52:43 +00:00
return {
x = tev.x ,
y = tev.y ,
id = tev.id ,
slot = tev.slot ,
2022-09-04 00:38:27 +00:00
timev = tev.timev , -- A ref is enough for this table, it's re-assigned to a new object on every SYN_REPORT
2014-03-13 13:52:43 +00:00
}
2012-12-30 05:44:55 +00:00
end
2012-11-16 00:58:01 +00:00
--[[
tap2 is the later tap
2012-12-30 05:44:55 +00:00
--]]
2020-10-28 21:47:42 +00:00
function GestureDetector : isTapBounce ( tap1 , tap2 , interval )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- NOTE: If time went backwards, make the delta infinite to avoid misdetections,
-- as we can no longer compute a sensible value...
2022-05-05 19:00:22 +00:00
local time_diff = tap2.timev - tap1.timev
if time_diff < 0 then
time_diff = time.huge
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-09-04 00:38:27 +00:00
if time_diff < interval then
local x_diff = math.abs ( tap1.x - tap2.x )
local y_diff = math.abs ( tap1.y - tap2.y )
return (
x_diff < self.SINGLE_TAP_BOUNCE_DISTANCE and
y_diff < self.SINGLE_TAP_BOUNCE_DISTANCE
)
end
2020-10-18 13:57:40 +00:00
end
2012-11-16 00:58:01 +00:00
function GestureDetector : isDoubleTap ( tap1 , tap2 )
2022-05-05 19:00:22 +00:00
local time_diff = tap2.timev - tap1.timev
if time_diff < 0 then
time_diff = time.huge
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-09-04 00:38:27 +00:00
if time_diff < self.ges_double_tap_interval then
local x_diff = math.abs ( tap1.x - tap2.x )
local y_diff = math.abs ( tap1.y - tap2.y )
return (
x_diff < self.DOUBLE_TAP_DISTANCE and
y_diff < self.DOUBLE_TAP_DISTANCE
)
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
end
2022-09-04 00:38:27 +00:00
function Contact : isTwoFingerTap ( buddy_contact )
local gesture_detector = self.ges_dec
2021-04-01 02:30:48 +00:00
2022-09-04 00:38:27 +00:00
local time_diff0 = self.current_tev . timev - self.initial_tev . timev
2022-05-05 19:00:22 +00:00
if time_diff0 < 0 then
time_diff0 = time.huge
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-09-04 00:38:27 +00:00
local time_diff1 = buddy_contact.current_tev . timev - buddy_contact.initial_tev . timev
2022-05-05 19:00:22 +00:00
if time_diff1 < 0 then
time_diff1 = time.huge
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-09-04 00:38:27 +00:00
if time_diff0 < gesture_detector.ges_two_finger_tap_duration and
time_diff1 < gesture_detector.ges_two_finger_tap_duration then
local x_diff0 = math.abs ( self.current_tev . x - self.initial_tev . x )
local x_diff1 = math.abs ( buddy_contact.current_tev . x - buddy_contact.initial_tev . x )
local y_diff0 = math.abs ( self.current_tev . y - self.initial_tev . y )
local y_diff1 = math.abs ( buddy_contact.current_tev . y - buddy_contact.initial_tev . y )
return (
x_diff0 < gesture_detector.TWO_FINGER_TAP_REGION and
x_diff1 < gesture_detector.TWO_FINGER_TAP_REGION and
y_diff0 < gesture_detector.TWO_FINGER_TAP_REGION and
y_diff1 < gesture_detector.TWO_FINGER_TAP_REGION
)
end
2013-03-06 13:04:33 +00:00
end
2017-04-12 16:55:22 +00:00
--[[--
2022-09-04 00:38:27 +00:00
Compares ` current_tev ` with ` initial_tev ` .
2017-04-12 16:55:22 +00:00
2022-09-04 00:38:27 +00:00
The first boolean argument ` simple ` results in only four directions if true .
2019-02-17 20:09:09 +00:00
2017-04-12 16:55:22 +00:00
@ return ( direction , distance ) pan direction and distance
2012-12-30 05:44:55 +00:00
--]]
2022-09-04 00:38:27 +00:00
function Contact : getPath ( simple , diagonal , initial_tev )
initial_tev = initial_tev or self.initial_tev
2019-02-17 20:09:09 +00:00
2022-09-04 00:38:27 +00:00
local x_diff = self.current_tev . x - initial_tev.x
local y_diff = self.current_tev . y - initial_tev.y
2014-03-13 13:52:43 +00:00
local direction = nil
local distance = math.sqrt ( x_diff * x_diff + y_diff * y_diff )
2016-09-12 04:47:24 +00:00
if x_diff ~= 0 or y_diff ~= 0 then
2014-03-13 13:52:43 +00:00
local v_direction = y_diff < 0 and " north " or " south "
local h_direction = x_diff < 0 and " west " or " east "
2019-03-02 12:22:44 +00:00
if ( not simple
and math.abs ( y_diff ) > 0.577 * math.abs ( x_diff )
and math.abs ( y_diff ) < 1.732 * math.abs ( x_diff ) )
or ( simple and diagonal )
2019-02-17 20:09:09 +00:00
then
2022-09-04 00:38:27 +00:00
direction = v_direction .. h_direction
2014-03-13 13:52:43 +00:00
elseif ( math.abs ( x_diff ) > math.abs ( y_diff ) ) then
direction = h_direction
else
direction = v_direction
end
end
return direction , distance
2013-03-03 14:18:38 +00:00
end
2022-09-04 00:38:27 +00:00
function Contact : isSwipe ( )
local gesture_detector = self.ges_dec
local time_diff = self.current_tev . timev - self.initial_tev . timev
2022-05-05 19:00:22 +00:00
if time_diff < 0 then
time_diff = time.huge
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-09-04 00:38:27 +00:00
if time_diff < gesture_detector.ges_swipe_interval then
local x_diff = self.current_tev . x - self.initial_tev . x
local y_diff = self.current_tev . y - self.initial_tev . y
2014-03-13 13:52:43 +00:00
if x_diff ~= 0 or y_diff ~= 0 then
return true
end
end
2012-12-30 05:44:55 +00:00
end
2013-03-28 13:06:13 +00:00
function GestureDetector : getRotate ( orig_point , start_point , end_point )
2022-09-04 00:38:27 +00:00
--[[
2014-03-13 13:52:43 +00:00
local a = orig_point : distance ( start_point )
local b = orig_point : distance ( end_point )
local c = start_point : distance ( end_point )
return math.acos ( ( a * a + b * b - c * c ) / ( 2 * a * b ) ) * 180 / math.pi
2022-09-04 00:38:27 +00:00
--]]
-- NOTE: I am severely maths impaired, and I just wanted something that preserved rotation direction (CCW if < 0),
-- so this is shamelessly stolen from https://stackoverflow.com/a/31334882
-- & https://stackoverflow.com/a/21484228
local rad = math.atan2 ( end_point.y - orig_point.y , end_point.x - orig_point.x ) -
math.atan2 ( start_point.y - orig_point.y , start_point.x - orig_point.x )
-- Normalize to [-180, 180]
if rad < - math.pi then
rad = rad + 2 * math.pi
elseif rad > math.pi then
rad = rad - 2 * math.pi
end
return rad * 180 / math.pi
2013-03-28 13:06:13 +00:00
end
2022-09-04 00:38:27 +00:00
function Contact : switchState ( state_func , func_arg )
self.state = state_func
return state_func ( self , func_arg )
2012-11-16 00:58:01 +00:00
end
2022-09-04 00:38:27 +00:00
function Contact : initialState ( )
local tev = self.current_tev
2013-03-26 12:59:47 +00:00
2014-03-13 13:52:43 +00:00
if tev.id then
2022-09-04 00:38:27 +00:00
-- Contact lift
2014-03-13 13:52:43 +00:00
if tev.id == - 1 then
2022-09-04 00:38:27 +00:00
-- If this slot was a buddy slot that happened to be dropped by a MT gesture in the *same* input frame,
-- a lift might be the first thing we process here... We can safely drop it again.
-- Hover pen events are also good candidates for this.
logger.dbg ( " Contact:initialState Cancelled a gesture in slot " , self.slot )
self.ges_dec : dropContact ( self )
2014-03-13 13:52:43 +00:00
else
2022-09-04 00:38:27 +00:00
self.id = tev.id
2014-03-13 13:52:43 +00:00
if tev.x and tev.y then
2022-09-04 00:38:27 +00:00
-- Contact down, user started a new touch motion
if not self.down then
self.down = true
2021-04-14 23:00:28 +00:00
-- NOTE: We can't use a simple reference, because tev is actually Input's self.ev_slots[slot],
-- and *that* is a fixed reference for a given slot!
2022-09-04 00:38:27 +00:00
-- Here, we really want to remember the *first* tev, so, make a copy of it.
self.initial_tev = deepCopyEv ( tev )
-- Default to tap state, indicating that this is a new contact
return self : switchState ( Contact.tapState , true )
2014-03-13 13:52:43 +00:00
end
end
end
end
2012-11-11 06:00:52 +00:00
end
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
--[[--
Attempts to figure out which clock source tap events are using ...
] ]
function GestureDetector : probeClockSource ( timev )
-- We'll check if that timestamp is +/- 2.5s away from the three potential clock sources supported by evdev.
-- We have bigger issues than this if we're parsing events more than 3s late ;).
2022-05-05 19:00:22 +00:00
local threshold = time.s ( 2 ) + time.ms ( 500 )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- Start w/ REALTIME, because it's the easiest to detect ;).
2022-05-05 19:00:22 +00:00
local realtime = time.realtime_coarse ( )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- clock-threshold <= timev <= clock+threshold
if timev >= realtime - threshold and timev <= realtime + threshold then
self.clock_id = C.CLOCK_REALTIME
2022-05-29 01:05:15 +00:00
logger.dbg ( " GestureDetector:probeClockSource: Touch event timestamps appear to use CLOCK_REALTIME " )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
return
end
-- Then MONOTONIC, as it's (hopefully) more common than BOOTTIME (and also guaranteed to be an usable clock source)
2022-05-05 19:00:22 +00:00
local monotonic = time.monotonic_coarse ( )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
if timev >= monotonic - threshold and timev <= monotonic + threshold then
self.clock_id = C.CLOCK_MONOTONIC
2022-05-29 01:05:15 +00:00
logger.dbg ( " GestureDetector:probeClockSource: Touch event timestamps appear to use CLOCK_MONOTONIC " )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
return
end
-- Finally, BOOTTIME
2022-05-05 19:00:22 +00:00
local boottime = time.boottime ( )
2021-04-13 15:53:15 +00:00
-- NOTE: It was implemented in Linux 2.6.39, so, reject 0, which would mean it's unsupported...
2022-05-05 19:00:22 +00:00
if not boottime == 0 and timev >= boottime - threshold and timev <= boottime + threshold then
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
self.clock_id = C.CLOCK_BOOTTIME
2022-05-29 01:05:15 +00:00
logger.dbg ( " GestureDetector:probeClockSource: Touch event timestamps appear to use CLOCK_BOOTTIME " )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
return
end
-- If we're here, the detection was inconclusive :/
self.clock_id = - 1
2022-05-29 01:05:15 +00:00
logger.dbg ( " GestureDetector:probeClockSource: Touch event clock source detection was inconclusive " )
2021-04-13 15:53:15 +00:00
-- Print all all the gory details in debug mode when this happens...
2022-05-05 19:00:22 +00:00
logger.dbg ( " Input frame : " , time.format_time ( timev ) )
logger.dbg ( " CLOCK_REALTIME : " , time.format_time ( realtime ) )
logger.dbg ( " CLOCK_MONOTONIC: " , time.format_time ( monotonic ) )
logger.dbg ( " CLOCK_BOOTTIME : " , time.format_time ( boottime ) )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
function GestureDetector : getClockSource ( )
return self.clock_id
end
function GestureDetector : resetClockSource ( )
self.clock_id = nil
end
2017-04-12 16:55:22 +00:00
--[[--
2022-09-04 00:38:27 +00:00
Handles both single and double tap . ` new_tap ` is true for the initial contact down event .
2012-11-17 18:41:26 +00:00
--]]
2022-09-04 00:38:27 +00:00
function Contact : tapState ( new_tap )
local slot = self.slot
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local gesture_detector = self.ges_dec
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
-- Attempt to detect the clock source for these events (we reset it on suspend to discriminate MONOTONIC from BOOTTIME).
2022-09-04 00:38:27 +00:00
if not gesture_detector.clock_id then
gesture_detector : probeClockSource ( tev.timev )
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415)
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
2021-03-30 00:57:59 +00:00
end
2022-08-11 15:56:40 +00:00
logger.dbg ( " slot " , slot , " in tap state... " )
2022-09-04 00:38:27 +00:00
-- Contact lift
2014-03-13 13:52:43 +00:00
if tev.id == - 1 then
2022-09-04 00:38:27 +00:00
if buddy_contact and self.down then
-- Both main contacts are actives and we are down
if self : isTwoFingerTap ( buddy_contact ) then
-- Mark that slot
self.mt_gesture = " tap "
-- Neuter its buddy
buddy_contact.state = Contact.voidState
buddy_contact.mt_gesture = " tap "
2014-03-13 13:52:43 +00:00
local pos0 = Geom : new {
2022-09-04 00:38:27 +00:00
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
local pos1 = Geom : new {
2022-09-04 00:38:27 +00:00
x = buddy_contact.current_tev . x ,
y = buddy_contact.current_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
local tap_span = pos0 : distance ( pos1 )
2022-09-04 21:38:13 +00:00
local tap_pos = pos0 : midpoint ( pos1 )
logger.dbg ( " two_finger_tap detected @ " , tap_pos.x , tap_pos.y , " with span " , tap_span )
2022-09-04 00:38:27 +00:00
-- Don't drop buddy, voidState will handle it
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
return {
ges = " two_finger_tap " ,
2022-09-04 21:38:13 +00:00
pos = tap_pos ,
2014-03-13 13:52:43 +00:00
span = tap_span ,
time = tev.timev ,
}
else
2022-09-04 21:38:13 +00:00
logger.dbg ( " Contact:tapState: Two-contact tap failed to pass the two_finger_tap constraints -> single tap @ " , tev.x , tev.y )
2022-09-04 00:38:27 +00:00
-- We blew the gesture position/time constraints,
-- neuter buddy and send a single tap on this slot.
buddy_contact.state = Contact.voidState
gesture_detector : dropContact ( self )
return {
ges = " tap " ,
pos = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
} ,
time = tev.timev ,
}
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
elseif self.down or self.pending_double_tap_timer then
-- Hand over to the double tap handler, it's responsible for downgrading to single tap
return self : handleDoubleTap ( )
2014-03-13 13:52:43 +00:00
else
2022-09-04 00:38:27 +00:00
-- Huh, caught a *second* contact lift for this contact? (should never happen).
logger.warn ( " Contact:tapState Cancelled a gesture " )
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
end
else
2022-09-04 00:38:27 +00:00
-- If we're pending a double_tap timer, flag the contact as down again.
if self.pending_double_tap_timer and self.down == false then
self.down = true
logger.dbg ( " Contact:tapState: Contact down " )
end
-- See if we need to do something with the move/hold
return self : handleNonTap ( new_tap )
2014-03-13 13:52:43 +00:00
end
2013-03-16 14:48:35 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits both tap & double_tap gestures . Contact is up ( but down is still true ) or pending a double_tap timer .
--]]
function Contact : handleDoubleTap ( )
local slot = self.slot
--logger.dbg("Contact:handleDoubleTap for slot", slot)
local tev = self.current_tev
local gesture_detector = self.ges_dec
-- If we don't actually detect two distinct taps (i.e., down -> up -> down -> up), then it's a hover, ignore it.
-- (Without a double tap timer involved, these get dropped in initialState).
if self.pending_double_tap_timer and self.down == false then
logger.dbg ( " Contact:handleDoubleTap Ignored a hover event " )
return
end
-- cur_tap is used for double tap and bounce detection
2014-03-13 13:52:43 +00:00
local cur_tap = {
x = tev.x ,
y = tev.y ,
timev = tev.timev ,
}
2022-09-04 00:38:27 +00:00
-- Tap interval / bounce detection may be tweaked by a widget (i.e., VirtualKeyboard)
local tap_interval = gesture_detector.input . tap_interval_override or gesture_detector.ges_tap_interval
-- We do tap bounce detection even when double tap is enabled
-- (so, double tap is triggered when: ges_tap_interval <= delay < ges_double_tap_interval).
if tap_interval ~= 0 and gesture_detector.previous_tap [ slot ] ~= nil and
gesture_detector : isTapBounce ( gesture_detector.previous_tap [ slot ] , cur_tap , tap_interval ) then
logger.dbg ( " Contact:handleDoubleTap Stopped a tap bounce " )
-- Simply ignore it, and drop this slot as this is a contact lift.
gesture_detector : dropContact ( self )
2020-10-18 13:57:40 +00:00
return
end
2022-09-04 00:38:27 +00:00
local ges_ev = {
-- Default to single tap
ges = " tap " ,
pos = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
} ,
time = tev.timev ,
}
if not gesture_detector.input . disable_double_tap and self.pending_double_tap_timer and
gesture_detector : isDoubleTap ( gesture_detector.previous_tap [ slot ] , cur_tap ) then
-- It is a double tap
2014-03-13 13:52:43 +00:00
ges_ev.ges = " double_tap "
2022-09-04 00:38:27 +00:00
logger.dbg ( " Contact:handleDoubleTap: double_tap detected @ " , ges_ev.pos . x , ges_ev.pos . y )
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
return ges_ev
end
2022-09-04 00:38:27 +00:00
-- Remember this tap, now that we're out of the bounce & double_tap windows
gesture_detector.previous_tap [ slot ] = cur_tap
2014-03-13 13:52:43 +00:00
2022-09-04 00:38:27 +00:00
if gesture_detector.input . disable_double_tap then
-- We can send the event immediately (no need for the timer stuff needed for double tap support)
logger.dbg ( " Contact:handleDoubleTap: single tap detected @ " , ges_ev.pos . x , ges_ev.pos . y )
gesture_detector : dropContact ( self )
2020-10-18 13:57:40 +00:00
return ges_ev
end
2022-09-04 00:38:27 +00:00
-- Double tap enabled: we can't send this single tap immediately as it may be the start of a double tap.
-- We'll send it as a single tap after a timer if no second tap happened in the double tap delay.
if not self.pending_double_tap_timer then
logger.dbg ( " set up double_tap timer " )
self.pending_double_tap_timer = true
-- setTimeout will handle computing the deadline in the least lossy way possible given the platform.
gesture_detector.input : setTimeout ( slot , " double_tap " , function ( )
if self == gesture_detector : getContact ( slot ) and self.pending_double_tap_timer then
self.pending_double_tap_timer = false
if self.state == Contact.tapState then
-- A single or double tap will yield a different contact object, by virtue of dropContact and closure magic ;).
-- Speaking of closures, this is the original ges_ev from the timer setup.
logger.dbg ( " double_tap timer detected a single tap in slot " , slot , " @ " , ges_ev.pos . x , ges_ev.pos . y )
gesture_detector : dropContact ( self )
return ges_ev
end
end
end , tev.timev , gesture_detector.ges_double_tap_interval )
end
-- Regardless of the timer shenanigans, it's at the very least a contact lift,
-- but we can't quite call dropContact yet, as it would cancel the timer.
self.down = false
logger.dbg ( " Contact:handleDoubleTap: Contact lift " )
end
--[[--
Handles move ( switch to panState ) & hold ( switch to holdState ) . Contact is down .
` new_tap ` is true for the initial contact down event .
--]]
function Contact : handleNonTap ( new_tap )
local slot = self.slot
--logger.dbg("Contact:handleNonTap for slot", slot)
local tev = self.current_tev
local gesture_detector = self.ges_dec
-- If we haven't yet fired the hold timer, do so first and foremost, as hold_pan handling *requires* a hold.
-- We only do this on the first contact down.
if new_tap and not self.pending_hold_timer then
2016-12-29 08:10:38 +00:00
logger.dbg ( " set up hold timer " )
2022-09-04 00:38:27 +00:00
self.pending_hold_timer = true
gesture_detector.input : setTimeout ( slot , " hold " , function ( )
-- If this contact is still active & alive and its timer hasn't been cancelled,
-- (e.g., it hasn't gone through dropContact because of a contact lift yet),
-- then check that we're still in a stationary contact down state (i.e., tapState).
-- NOTE: We need to check that the current contact in this slot is *still* the same object first, because closure ;).
if self == gesture_detector : getContact ( slot ) and self.pending_hold_timer then
self.pending_hold_timer = nil
if self.state == Contact.tapState and self.down then
-- NOTE: If we happened to have moved enough, holdState will generate a hold_pan on the *next* event,
-- but for now, the initial hold is mandatory.
-- On the other hand, if we're *already* in pan state, we stay there and *never* switch to hold.
2022-09-04 21:38:13 +00:00
logger.dbg ( " hold timer tripped a switch to hold state in slot " , slot )
2022-09-04 00:38:27 +00:00
return self : switchState ( Contact.holdState , true )
end
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
end , tev.timev , gesture_detector.ges_hold_interval )
-- NOTE: We only generate touch *once*, on first contact down (at which point there's not enough history to trip a pan).
2014-03-13 13:52:43 +00:00
return {
ges = " touch " ,
pos = Geom : new {
x = tev.x ,
y = tev.y ,
2022-09-04 00:38:27 +00:00
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
} ,
time = tev.timev ,
}
else
2022-09-04 00:38:27 +00:00
-- Once the hold timer has been fired, we're free to see if we can switch to pan,
-- if the contact moved far enough on the X or Y axes...
if ( math.abs ( tev.x - self.initial_tev . x ) >= gesture_detector.PAN_THRESHOLD ) or
( math.abs ( tev.y - self.initial_tev . y ) >= gesture_detector.PAN_THRESHOLD ) then
return self : switchState ( Contact.panState )
2014-03-13 13:52:43 +00:00
end
end
2012-11-16 00:58:01 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Handles the full panel of pans & swipes , including their two - finger variants .
--]]
function Contact : panState ( keep_contact )
local slot = self.slot
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local gesture_detector = self.ges_dec
2022-08-11 15:56:40 +00:00
logger.dbg ( " slot " , slot , " in pan state... " )
2014-03-13 13:52:43 +00:00
if tev.id == - 1 then
2022-09-04 00:38:27 +00:00
-- End of pan, emit swipe and swipe-like gestures if necessary
if self : isSwipe ( ) then
if buddy_contact and self.down then
-- Both main contacts are actives and we are down, mark that slot
self.mt_gesture = " swipe "
-- Neuter its buddy
-- NOTE: Similar trickery as in handlePan to deal with rotate,
-- without the panState check for self, because we're obviously in panState...
if buddy_contact.state ~= Contact.panState and buddy_contact.mt_immobile then
buddy_contact.mt_gesture = " rotate "
else
buddy_contact.mt_gesture = " swipe "
end
buddy_contact.state = Contact.voidState
local ges_ev = self : handleTwoFingerPan ( buddy_contact )
2014-03-13 13:52:43 +00:00
if ges_ev then
2022-09-04 00:38:27 +00:00
if buddy_contact.mt_gesture == " swipe " then
-- Only accept gestures that require both contacts to have been lifted
if ges_ev.ges == " two_finger_pan " then
ges_ev.ges = " two_finger_swipe "
elseif ges_ev.ges == " inward_pan " then
ges_ev.ges = " pinch "
elseif ges_ev.ges == " outward_pan " then
ges_ev.ges = " spread "
else
ges_ev = nil
end
else
-- Only accept the rotate gesture
if ges_ev.ges ~= " rotate " then
ges_ev = nil
end
end
if ges_ev then
logger.dbg ( ges_ev.ges , ges_ev.direction , ges_ev.distance or math.abs ( ges_ev.angle ) , " detected " )
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
end
-- Don't drop buddy, voidState will handle it.
-- NOTE: This is a hack for out of order rotate lifts when we have to fake a lift from voidState:
-- when `keep_contact` is true, this isn't an actual contact lift,
-- so we can't destroy the contact just yet...
if not keep_contact then
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
end
return ges_ev
2022-09-04 00:38:27 +00:00
elseif self.down then
return self : handleSwipe ( )
end
elseif self.down then
-- If the contact lift is not a swipe, then it's a pan.
return self : handlePanRelease ( keep_contact )
else
-- Huh, caught a *second* contact lift for this contact? (should never happen).
logger.warn ( " Contact:panState Cancelled a gesture " )
gesture_detector : dropContact ( self )
end
else
return self : handlePan ( )
end
end
--[[--
Used to ignore a buddy slot part of a MT gesture , so that we don ' t send duplicate events.
--]]
function Contact : voidState ( )
local slot = self.slot
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local buddy_slot = buddy_contact and self.buddy_contact . slot
local gesture_detector = self.ges_dec
logger.dbg ( " slot " , slot , " in void state... " )
-- We basically don't do anything but drop the slot on contact lift,
-- if need be deferring to the right state when we're part of a MT gesture.
if tev.id == - 1 then
if self.down and buddy_contact and buddy_contact.down and self.mt_gesture then
-- If we were lifted before our buddy, and we're part of a MT gesture,
-- defer to the proper state (wthout switching state ourselves).
if self.mt_gesture == " tap " then
return self : tapState ( )
elseif self.mt_gesture == " swipe " or self.mt_gesture == " pan " or self.mt_gesture == " pan_release " then
return self : panState ( )
elseif self.mt_gesture == " rotate " then
-- NOTE: As usual, rotate requires some trickery,
-- because it's the only gesture that requires both slots to be in *different* states...
-- (The trigger contact *has* to be the panning one; while we're the held one in this scenario).
logger.dbg ( " Contact:voidState Deferring to panState via buddy slot " , buddy_slot , " to handle MT contact lift for a rotate " )
local ges_ev
local buddy_tid = buddy_contact.current_tev . id
if buddy_tid == - 1 then
-- It's an actual lift for buddy, so we can just send it along, panState will drop the contact.
ges_ev = buddy_contact : panState ( )
else
-- But *this* means the lifts are staggered, and we, the hold pivot, were lifted *first*.
-- To avoid further issues, we'll forcibly lift buddy for this single call,
-- to make sure panState tries for the rotate gesture *now*,
-- while asking it *not* to drop itself just now (as it's not an actual contact lift just yet) so that...
buddy_contact.current_tev . id = - 1
ges_ev = buddy_contact : panState ( true )
-- ...we can then send it to the void.
-- Whether the gesture fails or not, it'll be in voidState and only dropped on actual contact lift,
-- regardless of whether the driver repeats ABS_MT_TRACKING_ID values or not.
-- Otherwise, if it only lifts on the next input frame,
-- it won't go through MT codepaths at all, and you'll end up with a single swipe,
-- and if it lifts even later, we'd have to deal with spurious moves first, probably leading into a tap...
-- If the gesture *succeeds*, the buddy contact will be dropped whenever it's actually lifted,
-- thanks to the temporary tracking id switcheroo & voidState...
buddy_contact.state = Contact.voidState
buddy_contact.current_tev . id = buddy_tid
end
-- Regardless of whether we detected a gesture, this is a contact lift, so it's curtains for us!
gesture_detector : dropContact ( self )
return ges_ev
elseif self.mt_gesture == " hold " or self.mt_gesture == " hold_pan " or
self.mt_gesture == " hold_release " or self.mt_gesture == " hold_pan_release " then
return self : holdState ( )
2014-03-13 13:52:43 +00:00
else
2022-09-04 00:38:27 +00:00
-- Should absolutely never happen (and, at the time of writing, is technically guaranteed to be unreachable).
logger.warn ( " Contact:voidState Unknown MT gesture " , self.mt_gesture , " cannot handle contact lift properly " )
-- We're still gone, though.
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
elseif self.down then
-- We were lifted *after* our buddy, the gesture already went through, we can silently slink away into the night.
logger.dbg ( " Contact:voidState Contact lift detected " )
gesture_detector : dropContact ( self )
else
-- Huh, caught a *second* contact lift for this contact? (should never happen).
logger.warn ( " Contact:voidState Cancelled a gesture " )
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
end
else
2022-09-04 00:38:27 +00:00
-- We need to be able to discriminate between a moving and unmoving contact for rotate/pan discrimination.
if self.mt_immobile then
if ( math.abs ( tev.x - self.initial_tev . x ) >= gesture_detector.PAN_THRESHOLD ) or
( math.abs ( tev.y - self.initial_tev . y ) >= gesture_detector.PAN_THRESHOLD ) then
self.mt_immobile = false
-- NOTE: We've just moved: if we were flagged for a hold gesture (meaning our buddy is still in holdState),
-- that won't do anymore, switch to a setup for a rotate gesture.
-- (This happens when attempting a rotate, and the hold timer for slot 0 expires
-- before slot 1 has the chance to switch to panState).
-- A.K.A., "rotate dirty hack #42" ;).
if buddy_contact and buddy_contact.mt_gesture == " hold " and self.mt_gesture == " hold " then
self.mt_gesture = " pan "
buddy_contact.mt_gesture = " rotate "
logger.dbg ( " Contact:voidState We moved while pending a hold gesture, swap to a rotate setup " )
end
else
self.mt_immobile = true
end
2014-03-13 13:52:43 +00:00
end
end
2013-03-28 13:06:13 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits the swipe & multiswipe gestures . Contact is up . ST only ( i.e . , there isn ' t any buddy contact active).
--]]
function Contact : handleSwipe ( )
--logger.dbg("Contact:handleSwipe for slot", self.slot)
local tev = self.current_tev
local gesture_detector = self.ges_dec
local swipe_direction , swipe_distance = self : getPath ( )
2014-03-13 13:52:43 +00:00
local start_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = self.initial_tev . x ,
y = self.initial_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
2019-02-17 20:09:09 +00:00
local ges = " swipe "
local multiswipe_directions
if # self.multiswipe_directions > 1 then
ges = " multiswipe "
multiswipe_directions = " "
2021-08-03 18:57:53 +00:00
for k , v in ipairs ( self.multiswipe_directions ) do
2019-02-17 20:09:09 +00:00
local sep = " "
if k > 1 then
sep = " "
end
multiswipe_directions = multiswipe_directions .. sep .. v [ 1 ]
end
logger.dbg ( " multiswipe " , multiswipe_directions )
end
2022-09-04 00:38:27 +00:00
logger.dbg ( " Contact:handleSwipe: swipe " , swipe_direction , swipe_distance , " detected " )
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
return {
2019-02-17 20:09:09 +00:00
ges = ges ,
2014-03-13 13:52:43 +00:00
-- use first pan tev coordination as swipe start point
pos = start_pos ,
direction = swipe_direction ,
2019-02-17 20:09:09 +00:00
multiswipe_directions = multiswipe_directions ,
2014-03-13 13:52:43 +00:00
distance = swipe_distance ,
time = tev.timev ,
}
2013-03-28 13:06:13 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits the pan gestures and handles their two finger variants . Contact is down ( and either in holdState or panState ) .
--]]
function Contact : handlePan ( )
--logger.dbg("Contact:handlePan for slot", self.slot)
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local gesture_detector = self.ges_dec
if buddy_contact and self.down then
-- Both main contacts are actives and we are down, mark that slot
self.mt_gesture = " pan "
-- Neuter its buddy
-- NOTE: Small trickery for rotate, which requires both contacts to be in very specific states.
-- We merge tapState with holdState because it's likely that the hold hasn't taken yet,
-- and it never will after that because we switch to voidState ;).
if buddy_contact.state ~= Contact.panState and buddy_contact.mt_immobile and
self.state == Contact.panState then
buddy_contact.mt_gesture = " rotate "
else
buddy_contact.mt_gesture = " pan "
end
buddy_contact.state = Contact.voidState
return self : handleTwoFingerPan ( buddy_contact )
elseif self.down then
local pan_direction , pan_distance = self : getPath ( )
2014-03-13 13:52:43 +00:00
local pan_ev = {
ges = " pan " ,
relative = {
-- default to pan 0
x = 0 ,
y = 0 ,
} ,
pos = nil ,
direction = pan_direction ,
distance = pan_distance ,
time = tev.timev ,
}
2019-02-28 22:49:23 +00:00
2022-09-04 00:38:27 +00:00
-- Regular pan
pan_ev.relative . x = tev.x - self.initial_tev . x
pan_ev.relative . y = tev.y - self.initial_tev . y
2019-02-28 22:49:23 +00:00
2014-03-13 13:52:43 +00:00
pan_ev.pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
2019-02-17 20:09:09 +00:00
local msd_cnt = # self.multiswipe_directions
local msd_direction_prev = ( msd_cnt > 0 ) and self.multiswipe_directions [ msd_cnt ] [ 1 ] or " "
2022-09-04 00:38:27 +00:00
local prev_ms_ev , fake_initial_tev
2019-02-22 11:36:03 +00:00
2019-03-01 20:47:45 +00:00
if msd_cnt == 0 then
2019-03-02 12:22:44 +00:00
-- determine whether to initiate a straight or diagonal multiswipe
self.multiswipe_type = " straight "
2022-09-04 00:38:27 +00:00
if pan_direction ~= " north " and pan_direction ~= " south " and
pan_direction ~= " east " and pan_direction ~= " west " then
2019-03-02 12:22:44 +00:00
self.multiswipe_type = " diagonal "
2019-03-01 20:47:45 +00:00
end
elseif msd_cnt > 0 then
2022-09-04 00:38:27 +00:00
-- recompute a more accurate direction and distance in a multiswipe context
2019-02-22 11:36:03 +00:00
prev_ms_ev = self.multiswipe_directions [ msd_cnt ] [ 2 ]
2022-09-04 00:38:27 +00:00
fake_initial_tev = {
x = prev_ms_ev.pos . x ,
y = prev_ms_ev.pos . y ,
2019-02-22 11:36:03 +00:00
}
end
2019-02-17 20:09:09 +00:00
2022-09-04 00:38:27 +00:00
-- the first time, fake_initial_tev is nil, so the contact's initial_tev is automatically used instead
2019-03-02 12:22:44 +00:00
local msd_direction , msd_distance
if self.multiswipe_type == " straight " then
2022-09-04 00:38:27 +00:00
msd_direction , msd_distance = self : getPath ( true , false , fake_initial_tev )
2019-03-02 12:22:44 +00:00
else
2022-09-04 00:38:27 +00:00
msd_direction , msd_distance = self : getPath ( true , true , fake_initial_tev )
2019-03-02 12:22:44 +00:00
end
2019-02-23 20:34:03 +00:00
2022-09-04 00:38:27 +00:00
if msd_distance > gesture_detector.MULTISWIPE_THRESHOLD then
2019-03-04 18:01:01 +00:00
local pan_ev_multiswipe = pan_ev
2022-09-04 00:38:27 +00:00
-- store a copy of pan_ev without rotation adjustment for multiswipe calculations when rotated
2022-12-21 14:50:39 +00:00
if gesture_detector.screen : getTouchRotation ( ) > gesture_detector.screen . DEVICE_ROTATED_UPRIGHT then
2019-03-04 18:01:01 +00:00
pan_ev_multiswipe = util.tableDeepCopy ( pan_ev )
end
2019-02-23 20:34:03 +00:00
if msd_direction ~= msd_direction_prev then
self.multiswipe_directions [ msd_cnt + 1 ] = {
[ 1 ] = msd_direction ,
2019-03-04 18:01:01 +00:00
[ 2 ] = pan_ev_multiswipe ,
2019-02-23 20:34:03 +00:00
}
else
2022-09-04 00:38:27 +00:00
-- update ongoing swipe direction to the new maximum
2019-02-23 20:34:03 +00:00
self.multiswipe_directions [ msd_cnt ] = {
[ 1 ] = msd_direction ,
2019-03-04 18:01:01 +00:00
[ 2 ] = pan_ev_multiswipe ,
2019-02-23 20:34:03 +00:00
}
2019-02-17 20:09:09 +00:00
end
end
2014-03-13 13:52:43 +00:00
return pan_ev
end
2012-11-16 00:58:01 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits the pan , two_finger_pan , two_finger_hold_pan , inward_pan , outward_pan & rotate gestures .
Contact is down in panState or holdState , or up in panState if it was lifted below the swipe interval .
--]]
function Contact : handleTwoFingerPan ( buddy_contact )
--logger.dbg("Contact:handleTwoFingerPan for slot", self.slot)
local gesture_detector = self.ges_dec
-- triggering contact is self
-- reference contact is buddy_contact
local tpan_dir , tpan_dis = self : getPath ( )
2014-03-13 13:52:43 +00:00
local tstart_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = self.initial_tev . x ,
y = self.initial_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
local tend_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = self.current_tev . x ,
y = self.current_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
local rstart_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = buddy_contact.initial_tev . x ,
y = buddy_contact.initial_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
2022-09-04 00:38:27 +00:00
if self.current_tev . id == - 1 and buddy_contact.mt_gesture == " rotate " then
-- NOTE: We only handle the rotate gesture when triggered by the just lifted pan finger
-- (actually, it needs to pass the swipe interval check, but it is in panState),
-- because this gesture would be too difficult to discriminate from a pinch/spread the other way around ;).
-- TL;DR: Both fingers need to move for a pinch/spread, while a finger needs to stay still for a rotate.
-- NOTE: FWIW, on an Elipsa, if we misdetect a pinch (i.e., both fingers moved) for a rotate
-- because the buddy slot failed to pass the pan threshold, we get a very shallow angle (often < 1°, at most ~2°).
-- If, on the other hand, we misdetect a rotate that *looked* like a pinch,
-- (i.e., a pinch with only one finger moving), we get slightly larger angles (~5°).
-- Things get wildly more difficult on an Android phone, where you can easily add ~10° of noise to those results.
-- TL;DR: We just chuck those as misdetections instead of adding brittle heuristics to correct course ;).
local angle = gesture_detector : getRotate ( rstart_pos , tstart_pos , tend_pos )
logger.dbg ( " Contact:handleTwoFingerPan: rotate " , angle , " detected " )
return {
ges = " rotate " ,
pos = rstart_pos ,
angle = angle ,
direction = angle >= 0 and " cw " or " ccw " ,
time = self.current_tev . timev ,
}
else
local rpan_dir , rpan_dis = buddy_contact : getPath ( )
2014-03-13 13:52:43 +00:00
local rend_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = buddy_contact.current_tev . x ,
y = buddy_contact.current_tev . y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
2022-09-21 21:26:22 +00:00
-- Use midpoint of tstart and rstart as swipe start point
local start_point = tstart_pos : midpoint ( rstart_pos )
local end_point = tend_pos : midpoint ( rend_pos )
-- Compute the distance based on the start & end midpoints
local avg_distance = start_point : distance ( end_point )
-- We'll also want to remember the span between both contacts on start & end for some gestures
2014-03-13 13:52:43 +00:00
local start_distance = tstart_pos : distance ( rstart_pos )
local end_distance = tend_pos : distance ( rend_pos )
local ges_ev = {
ges = " two_finger_pan " ,
2022-09-21 21:26:22 +00:00
pos = start_point ,
distance = avg_distance ,
2014-03-13 13:52:43 +00:00
direction = tpan_dir ,
2022-09-04 00:38:27 +00:00
time = self.current_tev . timev ,
2014-03-13 13:52:43 +00:00
}
if tpan_dir ~= rpan_dir then
if start_distance > end_distance then
ges_ev.ges = " inward_pan "
else
ges_ev.ges = " outward_pan "
end
2022-09-04 00:38:27 +00:00
ges_ev.direction = gesture_detector.DIRECTION_TABLE [ tpan_dir ]
2022-09-21 21:26:22 +00:00
-- Use the the sum of both contacts' travel for the distance
ges_ev.distance = tpan_dis + rpan_dis
-- Some handlers might also want to know the distance between the two contacts on lift & down.
ges_ev.span = end_distance
ges_ev.start_span = start_distance
2022-09-04 00:38:27 +00:00
elseif self.state == Contact.holdState then
ges_ev.ges = " two_finger_hold_pan "
-- Flag 'em for holdState to discriminate with two_finger_hold_release
self.mt_gesture = " hold_pan "
buddy_contact.mt_gesture = " hold_pan "
2014-03-13 13:52:43 +00:00
end
2022-09-04 00:38:27 +00:00
logger.dbg ( " Contact:handleTwoFingerPan: " , ges_ev.ges , ges_ev.direction , ges_ev.distance , " detected " )
2014-03-13 13:52:43 +00:00
return ges_ev
end
2013-03-28 13:06:13 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits the pan_release & two_finger_pan_release gestures . Contact is up ( but down is still true ) and in panState .
--]]
function Contact : handlePanRelease ( keep_contact )
--logger.dbg("Contact:handlePanRelease for slot", self.slot)
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local gesture_detector = self.ges_dec
2014-03-13 13:52:43 +00:00
local release_pos = Geom : new {
2022-09-04 00:38:27 +00:00
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
2014-03-13 13:52:43 +00:00
}
local pan_ev = {
ges = " pan_release " ,
pos = release_pos ,
time = tev.timev ,
}
2022-09-04 00:38:27 +00:00
if buddy_contact and self.down then
-- Both main contacts are actives and we are down, mark that slot
self.mt_gesture = " pan_release "
-- Neuter its buddy
buddy_contact.state = Contact.voidState
buddy_contact.mt_gesture = " pan_release "
logger.dbg ( " Contact:handlePanRelease: two_finger_pan_release detected " )
2014-03-13 13:52:43 +00:00
pan_ev.ges = " two_finger_pan_release "
2022-09-04 00:38:27 +00:00
-- Don't drop buddy, voidState will handle it
-- NOTE: This is yet another rotate hack, emanating from voidState into panState.
if not keep_contact then
gesture_detector : dropContact ( self )
end
return pan_ev
elseif self.down then
logger.dbg ( " Contact:handlePanRelease: pan release detected " )
gesture_detector : dropContact ( self )
return pan_ev
2014-03-13 13:52:43 +00:00
else
2022-09-04 00:38:27 +00:00
-- Huh, caught a *second* contact lift for this contact? (should never happen).
logger.warn ( " Contact:handlePanRelease Cancelled a gesture " )
gesture_detector : dropContact ( self )
2014-03-13 13:52:43 +00:00
end
2013-03-28 13:06:13 +00:00
end
2022-09-04 00:38:27 +00:00
--[[--
Emits the hold , hold_release & hold_pan gestures and their two_finger variants .
--]]
function Contact : holdState ( new_hold )
local slot = self.slot
local tev = self.current_tev
local buddy_contact = self.buddy_contact
local gesture_detector = self.ges_dec
2022-08-11 15:56:40 +00:00
logger.dbg ( " slot " , slot , " in hold state... " )
2022-09-04 00:38:27 +00:00
-- When we switch to hold state, we pass an additional boolean param "new_hold".
if new_hold and tev.id ~= - 1 then
if buddy_contact and self.down then
-- Both main contacts are actives and we are down, mark that slot
self.mt_gesture = " hold "
-- Neuter its buddy
buddy_contact.state = Contact.voidState
buddy_contact.mt_gesture = " hold "
local pos0 = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
}
local pos1 = Geom : new {
x = buddy_contact.current_tev . x ,
y = buddy_contact.current_tev . y ,
w = 0 ,
h = 0 ,
}
local tap_span = pos0 : distance ( pos1 )
2022-09-04 21:38:13 +00:00
local tap_pos = pos0 : midpoint ( pos1 )
logger.dbg ( " two_finger_hold detected @ " , tap_pos.x , tap_pos.y , " with span " , tap_span )
2022-09-04 00:38:27 +00:00
return {
ges = " two_finger_hold " ,
2022-09-04 21:38:13 +00:00
pos = tap_pos ,
2022-09-04 00:38:27 +00:00
span = tap_span ,
time = tev.timev ,
}
elseif self.down then
2022-09-04 21:38:13 +00:00
logger.dbg ( " hold detected @ " , tev.x , tev.y )
2022-09-04 00:38:27 +00:00
return {
ges = " hold " ,
pos = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
} ,
time = tev.timev ,
}
end
elseif tev.id == - 1 then
if buddy_contact and self.down then
-- Both main contacts are actives and we are down, mark that slot
if self.mt_gesture == " rotate " and buddy_contact.mt_gesture == " pan " then
-- NOTE: We're setup as the hold in a rotate gesture, and we were lifted *before* our pan buddy,
-- do a bit of gymnastics, because the trigger contact for a rotate *needs* to be the pan...
-- This is a snow protocol special :/.
-- NOTE: This is simpler than the elaborate trickery this case involves when dealt with via voidState,
-- because it is specifically aimed at the snow protocol, so we *know* both contacts are lifted
-- in the same input frame.
logger.dbg ( " Contact:holdState: Early lift as a rotate pivot, trying for a rotate... " )
local ges_ev = buddy_contact : handleTwoFingerPan ( self )
if ges_ev then
if ges_ev.ges ~= " rotate " then
ges_ev = nil
else
logger.dbg ( ges_ev.ges , ges_ev.direction , math.abs ( ges_ev.angle ) , " detected " )
end
end
-- Regardless of whether this panned out (pun intended), this is a lift, so we'll defer to voidState next.
buddy_contact.state = Contact.voidState
gesture_detector : dropContact ( self )
return ges_ev
elseif self.mt_gesture == " hold_pan " or self.mt_gesture == " pan " then
self.mt_gesture = " hold_pan_release "
buddy_contact.mt_gesture = " hold_pan_release "
else
self.mt_gesture = " hold_release "
buddy_contact.mt_gesture = " hold_release "
end
-- Neuter its buddy
buddy_contact.state = Contact.voidState
-- Don't drop buddy, voidState will handle it
gesture_detector : dropContact ( self )
2022-09-04 21:38:13 +00:00
local pos0 = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
}
local pos1 = Geom : new {
x = buddy_contact.current_tev . x ,
y = buddy_contact.current_tev . y ,
w = 0 ,
h = 0 ,
}
local ges_type = self.mt_gesture == " hold_pan_release " and " two_finger_hold_pan_release " or " two_finger_hold_release "
local tap_span = pos0 : distance ( pos1 )
local tap_pos = pos0 : midpoint ( pos1 )
logger.dbg ( ges_type , " detected @ " , tap_pos.x , tap_pos.y , " with span " , tap_span )
2022-09-04 00:38:27 +00:00
return {
2022-09-04 21:38:13 +00:00
ges = ges_type ,
pos = tap_pos ,
span = tap_span ,
2022-09-04 00:38:27 +00:00
time = tev.timev ,
}
elseif self.down then
-- Contact lift, emit a hold_release
2022-09-04 21:38:13 +00:00
logger.dbg ( " hold_release detected @ " , tev.x , tev.y )
2022-09-04 00:38:27 +00:00
gesture_detector : dropContact ( self )
return {
ges = " hold_release " ,
pos = Geom : new {
x = tev.x ,
y = tev.y ,
w = 0 ,
h = 0 ,
} ,
time = tev.timev ,
}
else
-- Huh, caught a *second* contact lift for this contact? (should never happen).
logger.warn ( " Contact:holdState Cancelled a gesture " )
gesture_detector : dropContact ( self )
end
elseif tev.id ~= - 1 and ( ( math.abs ( tev.x - self.initial_tev . x ) >= gesture_detector.PAN_THRESHOLD ) or
( math.abs ( tev.y - self.initial_tev . y ) >= gesture_detector.PAN_THRESHOLD ) ) then
-- We've moved enough to count as a pan, defer to the pan handler, but stay in holdState
local ges_ev = self : handlePan ( )
if ges_ev ~= nil then
if ges_ev.ges ~= " two_finger_hold_pan " then
ges_ev.ges = " hold_pan "
end
end
2014-03-13 13:52:43 +00:00
return ges_ev
end
2012-11-13 06:12:01 +00:00
end
2012-11-11 06:00:52 +00:00
2020-07-09 17:11:44 +00:00
local ges_coordinate_translation_270 = {
2019-02-18 15:42:34 +00:00
north = " west " ,
south = " east " ,
east = " north " ,
west = " south " ,
northeast = " northwest " ,
2019-03-15 18:53:15 +00:00
northwest = " southwest " ,
2019-02-18 15:42:34 +00:00
southeast = " northeast " ,
southwest = " southeast " ,
}
local ges_coordinate_translation_180 = {
north = " south " ,
south = " north " ,
east = " west " ,
west = " east " ,
northeast = " southwest " ,
northwest = " southeast " ,
southeast = " northwest " ,
southwest = " northeast " ,
}
2020-07-09 17:11:44 +00:00
local ges_coordinate_translation_90 = {
2019-02-18 15:42:34 +00:00
north = " east " ,
south = " west " ,
east = " south " ,
west = " north " ,
northeast = " southeast " ,
northwest = " northeast " ,
southeast = " southwest " ,
southwest = " northwest " ,
}
local function translateGesDirCoordinate ( direction , translation_table )
return translation_table [ direction ]
end
local function translateMultiswipeGesDirCoordinate ( multiswipe_directions , translation_table )
2019-03-04 18:01:01 +00:00
return multiswipe_directions : gsub ( " %S+ " , translation_table )
2019-02-18 15:42:34 +00:00
end
2017-04-12 16:55:22 +00:00
--[[--
Changes gesture ' s `x` and `y` coordinates according to screen view mode.
2013-02-19 03:13:24 +00:00
@ param ges gesture that you want to adjust
@ return adjusted gesture .
--]]
function GestureDetector : adjustGesCoordinate ( ges )
2020-09-22 20:04:37 +00:00
local mode = self.screen : getTouchRotation ( )
2022-12-21 14:50:39 +00:00
if mode == self.screen . DEVICE_ROTATED_CLOCKWISE then
2020-07-09 17:11:44 +00:00
-- in landscape mode rotated 90
2014-03-13 13:52:43 +00:00
if ges.pos then
2014-10-30 18:42:18 +00:00
ges.pos . x , ges.pos . y = ( self.screen : getWidth ( ) - ges.pos . y ) , ( ges.pos . x )
2014-03-13 13:52:43 +00:00
end
if ges.ges == " swipe " or ges.ges == " pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " hold_pan "
2019-02-18 15:42:34 +00:00
or ges.ges == " multiswipe "
2014-03-13 13:52:43 +00:00
or ges.ges == " two_finger_swipe "
2019-02-18 15:42:34 +00:00
or ges.ges == " two_finger_pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " two_finger_hold_pan "
2019-02-18 15:42:34 +00:00
then
2020-07-09 17:11:44 +00:00
ges.direction = translateGesDirCoordinate ( ges.direction , ges_coordinate_translation_90 )
2019-02-18 15:42:34 +00:00
if ges.ges == " multiswipe " then
2020-07-09 17:11:44 +00:00
ges.multiswipe_directions = translateMultiswipeGesDirCoordinate ( ges.multiswipe_directions , ges_coordinate_translation_90 )
2022-10-27 04:48:49 +00:00
logger.dbg ( " GestureDetector: Landscape translation for multiswipe: " , ges.multiswipe_directions )
else
logger.dbg ( " GestureDetector: Landscape translation for ges: " , ges.ges , ges.direction )
2014-03-13 13:52:43 +00:00
end
if ges.relative then
ges.relative . x , ges.relative . y = - ges.relative . y , ges.relative . x
end
elseif ges.ges == " pinch " or ges.ges == " spread "
or ges.ges == " inward_pan "
or ges.ges == " outward_pan " then
if ges.direction == " horizontal " then
ges.direction = " vertical "
elseif ges.direction == " vertical " then
ges.direction = " horizontal "
end
2022-10-27 04:48:49 +00:00
logger.dbg ( " GestureDetector: Landscape translation for ges: " , ges.ges , ges.direction )
2014-03-13 13:52:43 +00:00
end
2022-12-21 14:50:39 +00:00
elseif mode == self.screen . DEVICE_ROTATED_COUNTER_CLOCKWISE then
2020-07-09 17:11:44 +00:00
-- in landscape mode rotated 270
2014-03-13 13:52:43 +00:00
if ges.pos then
2014-10-30 18:42:18 +00:00
ges.pos . x , ges.pos . y = ( ges.pos . y ) , ( self.screen : getHeight ( ) - ges.pos . x )
2014-03-13 13:52:43 +00:00
end
if ges.ges == " swipe " or ges.ges == " pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " hold_pan "
2019-02-18 15:42:34 +00:00
or ges.ges == " multiswipe "
2014-03-13 13:52:43 +00:00
or ges.ges == " two_finger_swipe "
2019-02-18 15:42:34 +00:00
or ges.ges == " two_finger_pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " two_finger_hold_pan "
2019-02-18 15:42:34 +00:00
then
2020-07-09 17:11:44 +00:00
ges.direction = translateGesDirCoordinate ( ges.direction , ges_coordinate_translation_270 )
2019-02-18 15:42:34 +00:00
if ges.ges == " multiswipe " then
2020-07-09 17:11:44 +00:00
ges.multiswipe_directions = translateMultiswipeGesDirCoordinate ( ges.multiswipe_directions , ges_coordinate_translation_270 )
2022-10-27 04:48:49 +00:00
logger.dbg ( " GestureDetector: Inverted landscape translation for multiswipe: " , ges.multiswipe_directions )
else
logger.dbg ( " GestureDetector: Inverted landscape translation for ges: " , ges.ges , ges.direction )
2014-03-13 13:52:43 +00:00
end
if ges.relative then
ges.relative . x , ges.relative . y = ges.relative . y , - ges.relative . x
end
elseif ges.ges == " pinch " or ges.ges == " spread "
or ges.ges == " inward_pan "
or ges.ges == " outward_pan " then
if ges.direction == " horizontal " then
ges.direction = " vertical "
elseif ges.direction == " vertical " then
ges.direction = " horizontal "
end
2022-10-27 04:48:49 +00:00
logger.dbg ( " GestureDetector: Inverted landscape translation for ges: " , ges.ges , ges.direction )
2014-03-13 13:52:43 +00:00
end
2022-12-21 14:50:39 +00:00
elseif mode == self.screen . DEVICE_ROTATED_UPSIDE_DOWN then
2016-08-02 09:21:27 +00:00
-- in portrait mode rotated 180
if ges.pos then
ges.pos . x , ges.pos . y = ( self.screen : getWidth ( ) - ges.pos . x ) , ( self.screen : getHeight ( ) - ges.pos . y )
end
if ges.ges == " swipe " or ges.ges == " pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " hold_pan "
2019-02-18 15:42:34 +00:00
or ges.ges == " multiswipe "
or ges.ges == " two_finger_swipe "
or ges.ges == " two_finger_pan "
2022-09-04 00:38:27 +00:00
or ges.ges == " two_finger_hold_pan "
2019-02-18 15:42:34 +00:00
then
ges.direction = translateGesDirCoordinate ( ges.direction , ges_coordinate_translation_180 )
if ges.ges == " multiswipe " then
ges.multiswipe_directions = translateMultiswipeGesDirCoordinate ( ges.multiswipe_directions , ges_coordinate_translation_180 )
2022-10-27 04:48:49 +00:00
logger.dbg ( " GestureDetector: Inverted portrait translation for multiswipe: " , ges.multiswipe_directions )
else
logger.dbg ( " GestureDetector: Inverted portrait translation for ges: " , ges.ges , ges.direction )
2016-08-02 09:21:27 +00:00
end
if ges.relative then
ges.relative . x , ges.relative . y = - ges.relative . x , - ges.relative . y
end
end
2022-10-27 04:48:49 +00:00
-- pinch/spread are unaffected
2014-03-13 13:52:43 +00:00
end
return ges
2013-02-19 03:13:24 +00:00
end
2013-10-18 20:38:07 +00:00
return GestureDetector