- Fix sorting folders when collate is "type", "size", "percentage":
folders are sorted by name now.
- Fix getting next file in folder when collate is "mixed files and
folders": returned nil when next item was a folder.
The path list will start with \n already, but an extra newline
won't hurt and this string is really weird (i.e., people will
naturally put a space or a newline in front of it).
Add a dedicated handler to handle mixed pen + panel, when panel sends both mt + st and stylus only st
Don't stomp on panel slots in the Wacom protocol handler
Get rid of a few superfluous nil guards
Return early in the EV_KEY handler for touch-related events (i.e., BTN_)
- AutoFrontlight plugin: update checkmark on toggling
- KeepAlive plugin: update checkmark on toggling
- ReaderPageMap: font size menu entry
- ReaderStatus, common_settings_menu_table: book status "read" -> "finished"
* Add thinspaces between d/h/m/s
* Remove lead zeros
* Make letter `d` for days translatable
* Make thinspace into hairspace when compact
* Adjust and add unit tests
Fix some of my early blunders in using the `N_()` gettext function. Mini-PR from https://github.com/koreader/koreader/pull/9924#discussion_r1104298501 (@Frenzie).
There was also one line for generating this same `%1 (%2 pages)` text that confusingly uses different ordering in the SQL query output; switched the two SELECT arguments around to make it match the other 5 usages. Works the same as before
Some devices don't ship with iproute2 compatible tools, but may ship
with net-tools compatible ones.
Unify code w/ SDL's similar codepaths for its NetworkMgr:isConnected
implementation
- New dogear icons in Mosaic cover view to indicate each
possible book status: 'Reading', 'On hold', 'Finished'
- Progress bar redesigned to be floating, taller, thicker
bordered in Mosaic cover view
- Don't show progress bar if book is finished
- FakeCover bottom text (filename) adjusted to not overlap
with progress bar and dogear if they exist
- Mosaic book shortcut letter moved from bottom left to
top left
Only available with EPUBs containing 2 or more fragments,
and a file size large enough to ensure a cache file is used.
The idea is simply, on any rendering setting change, to
skip the rerendering of the full book and to defer any
rerendering to the moment we draw a DocFragment, and
render only it.
So, on a setting change, only the fragment containing the
current page will be rerendered, and the new fragments we
may cross while turning pages.
When having done so, KOReader is in a degraded state (the
full page count is incorrect, the ToC is invalid...).
So, a full rerendering is needed, and one will happen
in the background, and when the user is idle, we reload
seamlessly and quickly from the cache file it has made.
ReaderFlipping will show some icons in the top left
corner to let it know at which steps in this procress
we are.
Includes:
- Cache: minor simplifications
- Cache: improve cache file reproducibility
- CHM: minor cleanup
- LVStream: minor cleanups
- LVStream: fix and improve `LVZipDecodeStream` implementation
- LVStream: disable CRC checks for ZIP streams
- CSS: optimize :nth-child/of-type() and :nth-last-child/of-type()
- Page splitting: fix oversight in flags handling
- Update _doc_rendering_hash also when loading from cache
- getDocumentRenderingHash(extended): add parameter
- Toggable support for partial rerenderings of individual DocFragments
Also:
- util.isSubProcessDone(): add 'wait' option
- Also add getppid and some contants to ffi/posix.
Also don't propagate tap/hold on disabled menu items.
(When top and bottom menu displayed at the same time,
tap on a disabled top menu item could active some
bottom toggle change.)
Re #10124
----
Also includes:
* Resize the Zoom-to-Page icon to match the size of its brethren
It was one pixel smaller on all sides, for... some reason?
(Possibly I borked the resize the first time around, don't remember if I
worked on that one).
The previous iteration, besides failing to handle leaf-only input,
was relying on splitFilePathName, which just doesn't do what's required
to incrementally build the directory tree the right way around ;).
This should more closely match mkdir -p, i.e., it will *fail* if any
part (or all of it) of the path exists but is *not* a directory.
Re #10074
hius07
Frans de Jonge
zwim
NiLuJe
melyux
poire-z
ichnilatis-gr
Andrej Shadura
yparitcher