Rather than trying to force things in via library constructors
and weak symbols and --whole-archive and pkg-config tricks, just
add new functions ncdirect_core_init() and notcurses_core_init().
libnotcurses has ncdirect_init() and notcurses_init(), which pass
through to these. apps linking against notcurses-core ought use
the _core_ variants directly. This gets exactly the linkage we
want, everywhere. Convert many PoCs to _core_ variants. #1301
Signed-off-by: nick black <dankamongmen@gmail.com>
Extract `libnotcurses-core` from `libnotcurses`. The former contains everything except multimedia code. The latter contains multimedia stuff (a wrapper around FFmpeg or OIIO). If built with `-DUSE_MULTIMEDIA=none`, there will not be any `libnotcurses.so` generated. `libnotcurses.so` uses library constructors/destructors to insert its implementation into the `ncvisual` stack at runtime. Users linking `-lnotcurses` will get the full implementation; users linking `-lnotcurses-core` only will get the stack less multimedia code.
The upshot of this is that someone can compile/install only `libnotcurses-core`, and a program linked against it will work just fine. This eliminates the need to install the full (large) dependency stack of the multimedia code unless necessary. This will hopefully be useful for e.g. installers etc. Closes#339.
I've created the new files src/compat/compat.{hc}. These
are available to all binaries by adding src to the include
directories, and src/compat/compat.c to the sources. Several
functions are implemented here which one or more target
operating systems are missing, right now all related to time.
This includes clock_nanosleep(), which is missing on OS X and
DragonFly BSD. Eliminate the other three definitions of
timespec_to_ns() and friends. Standardize on NANOSECS_IN_SEC
rather than the more opaque GIG. Progress on #1121.
Remove the pre-rendered model.png, a generated output. Add a
dot target to CMake, predicated on USE_DOXYGEN (since we have
REQUIRED dot in our Doxygen spec). Generate the png in the
output path. Rename stacks to piles in model.dot.
Partially reverts: c5c9432ac0
Context: https://github.com/dankamongmen/notcurses/pull/1146
The above commit implemented the much more convenient way of finding and
using a package from CMake, however it broke compatibility with the way
id had been before, which defined various `Notcurses_*` variables.
Revert the `Notcurses` component to the old way and add support for
`Notcurses++` in the same fashion.
Perhaps one day we should think of implementing `find_package` support
by way of a module, which should make it possible to have the cake and
eat the cake. One day.
CMake:
Simplify cmake target+version config generation and make it actually
work. With the changes it is now possible to detect and use
`Notcurses` in the following way:
find_package(Notcurses REQUIRED)
...
target_link_libraries(myapp PRIVATE notcurses::notcurses)
Also, added the same CMake configuration for `Notcurses++`, to be used
in the following way:
find_package(Notcurses REQUIRED
find_package(Notcurses++ REQUIRED)
...
target_link_libraries(myapp PRIVATE notcurses++::notcurses++)
Docs:
`notcurses_cell(3)`: `cell_styles_{on,off} -> cell_{on,off}_styles`
and `cell_load_simple` -> `cell_load_char`
C++ API:
* Plane: added constructors taking `ncplane_options const&` instead of
the multitude of individual parameters
* Plane: drop `struct` when `ncplane_options` is used.
* Plane: added `strdup` (`cell_strdup`)
* Plane: added `extract` (`cell_extract`)
New reel layout algorithm based on trimming and sifting. Fixes the original issue of #818, though I'm not marking that bug fixed until I've resolved the little issues remaining with this one.
Back off CMake version dependency, see if we can get by with 3.11.4 for EPEL8 #851
Simplify tablet drawing tremendously by separating tablet border and data planes. Callbacks no longer need worry about the borders; they can simply fill the plane they're handed. #833
Improve notcurses_debug() a bit
Add ncplane_new_named() and friends to expose plane naming to the user.
Add internal ncplane_genocide() to kill a plane and all its bound descendents
New industrial-strength ncreel unit testing
notcurses-ncreel now accepts -ln for log level n
Add ncplane_parent() and ncplane_parent_const()
If we're fed a word that in and of itself is wider than
the text layout area, and it is not the first word on the
row on which it starts, go ahead and start it there, as
opposed to breaking the line. #720
Also, remove sgr from test suite, as it's not a notcurses
application.
Add back the USE_DOCTEST CMake option, but make it conditional
upon BUILD_TESTING. If BUILD_TESTING is provided without
USE_DOCTEST, we build and run the PoC tests, just not
notcurses-tester. If neither is defined, no tests are available
(the PoC binaries are still built). If both are defined, build
the PoCs plus notcurses-tester, and run them all in make test.
* Some things go to the FILE* we're provided. Some can only go to a controlling terminal. Check to see if the FILE we're given is a TTY. If not, open up /dev/tty #752.
* term_emit(): fflush() can return EAGAIN. Loop on it to eliminate a rare error on shutdown that particularly affected unit tests (where we start and shut down Notcurses many times in a row).
* sgr poc: check return value of setlocale()
* drone: run all unit tests
* CMake: add some tests using PoCs
* ncneofetch: print even small palettes
Fixes: https://github.com/dankamongmen/notcurses/issues/722
`-Wpedantic` is dropped because C code uses designated structure
initializers which are available in C++20 onward (and we use C++17
ATM). Once we switch to C++20 we should re-enable the flag.
`-fno-exceptions` is dropped because we do have exceptions, so...
This represents an essentially complete rewrite of ncvisual and associated code. It had two major goals:
Improve the ncvisual API based off lessons learned, pursuant to the upcoming API freeze. In particular, I wanted to:
decouple ncvisuals from ncplanes. It should be possible to render a ncvisual to multiple planes, with different scaling each time. It should be possible to create an ncvisual without a plane, etc.
normalize the various ways of constructing an ncvisual -- file, memory, plane, etc.
Support multiple blitters, from 7-bit ASCII to Sixel. This required writing the blitters in several cases, and they're not yet in their final implementations (but the API is fine)
I have not yet unified Plots and Visuals, and might not, given that the Plot code works fine. We could at this point implement Plots in terms of Visuals, though -- the blitter backend range has been unified. Sixel is not yet implemented, though it is listed.
There is a new POC tool, blitter. It renders its arguments using all possible blitter+scaling combinations. Another new POC, resize, displays its argument, then resizes it to the screen size and displays that, explicitly making use of ncvisual_resize() rather than a scaling parameter to ncvisual_render().
This also eliminates some memory leaks and bugs we were seeing in trunk, and brings in Sixel scaffolding.
The C++ wrapper will also need patching back up; I cut most of it down while wrestling with this crap, urk.
Closes#638, #562, and #622.
This commit introduces the same shared library versioning scheme as used
by the SDL library. The advantage is that different versions of
notcurses can be installed alongside each other (which is not an
unlikely scenario, as SDL itself certifies) and that, if the versioning
protocol is followed, any change to ABI will produce a DSO whose name
will not break any applications linked against any previous version.
Fixes: https://github.com/dankamongmen/notcurses/issues/616
SIGSEGV was caused by an invalid cast.
Short explanation: PEBKAC
Long explanation: `Selector.hh`, `Plot.hh` and `MultiSelector.hh` did
not include `Plane.hh`, they merely declared `class Plane;` because
inclusion of `Plane.hh` would cause circular dependencies to appear and
the compiler would be unhappy. On top of that, yours truly wrenched the
compiler's hands and caused it to believe that a pointer to `Plane` is
really a pointer to `ncplane*` which was quite a silly thing to do as
the compiler, not having included `Plane.hh` and thus not knowing full
definition of the type, wasn't able to look up the type cast operator in
`Plane`.
Don't abuse `reinterpret_cast`, kids!
