This is nothing more than duct tape, do not pull.
In the long run boost::filesystem should just be replaced with
std::filesystem (as done in tgoyne/Aegisub:cmake), but this would
complicate the history and cause merge conflicts for a bunch of pulls.
Hence this horrible temporary solution.
When Avisynth is not installed or not functional, this would otherwise
cause a crash when trying to initialize Avisynth more than once, since
after the first time the refcount would have been incrased anyway.
Also, allow negative fsp.
This doesn't change the behaviour of the SpinCtrl increment/decrement
buttons, just the precision that can manually be set. Small values of
fsp can be useful as a hack to disable all ligatures. Precise values of
fscx/fscy can be useful to compensate for anamorphic resolutions. The
other fields were made more precise for consistency.
This fixes a crash on Windows when double-clicking the draggable
separator between the column headers "Command" and "Description" in the
hotkey configuration dialog.
This makes the internal juggling of fold data even more complicated, but
it has a number of advantages:
- Folds will preserved even when opening the file with Aegisub builds
that don't know about folding.
- Folds will be preserved by automation scripts that re-insert every
line.
- Folds do not break when adding or deleting lines in a text editor or
some other editor, as long as the lines involving folds aren't
touched.
- In particular, merging changes using tools like git will not break
folds.
- Copy/pasting folds also keeps the folds to some extent. This is
impossible to solve in full generality, but simple cases like
cut/pasting a folded section somewhere else work, as long as the file
isn't saved in between.
- This will give automation scripts full access to folding without
needing to set up any additional API functions.
Unlike the other arguments for the resize function, "range_in" does not
use the same format as in the frame props. A frame prop _Range=1 means
limited, while an argument range_in=1 means full range.
The highlighting distinguishes drawing commands from coordinates, and
colors x and y coordinates in different colors to make coordinates
easier to visually parse. Furthermore, in cubic Bezier curves, it
underlines the coordinates which corresponds to endpoints of the curves.
Aegisub will automatically override the YCbCr Matrix field of the
current file's Script Properties with the video's reported color space.
The FFMS2 provider guesses a color space for all videos, but we don't
do this for Vapoursynth. Thus, we now disable this overriding whenever
the colorspace isn't known.
On Windows, Python changes the application's locale upon being called,
which will break wxwidgets, causing various assertion error dialogs or
even crashes (for example when interacting with any sort of float edit
control). Saving the locale and restoring it afterwards seems to be
the only really possible way to remedy this.
The percent values used for the overscan masks follow the BBC's
guidelines, as in
https://en.wikipedia.org/wiki/Overscan#Overscan_amounts .
However, these measure the per-side width as opposed to the total
percentage of width/height being cut off. Thus, they should not be
divided by two when drawing the mask.
The secondary_editor is a wxTextCtrl, whose height is calculated from 2 rows of text. Using this height gives better consistency on screens with different DPIs, instead of using hard coded value like 50px
Since FileNotFound exceptions don't abort the provider search, opening
an invalid path would show errors such as "vapoursynth not found" when
VapourSynth wasn't installed, even if VapourSynth wasn't selected as the
default video provider.
This default script will be executed to load any file whose file name
extension is not .py or .vpy .
The gui code for setting the default script is still a bit wonky as it
doesn't fit the rest of the preferences pages nicely, but it works for
now.
Since FileNotFound exceptions don't abort the provider search, opening
an invalid path would show errors such as "avisynth not found" when
Avisynth wasn't installed, even if Avisynth wasn't selected as the video
provider.
When cancelling an automation macro from the progress dialog, the dialog
throws a UserCancelException. If the macro still runs to the end
afterwards (instead of calling aegisub.cancel or causing an exception),
the two return values are left on the stack. This causes assertion errors
due to check_stack when those are enabled.
Some of these got lost in the process before.
For now, default to 709 if the colorspace is unkown. More detailed
colorspace guessing can be added in the future.
These alignment flags would cause an assertion error
"Horizontal alignment flags are ignored in horizontal sizers"
when opening the dialog (mainly when importing styles from another
script).
As the assertion error says, the flags are ignored anyway, so they're
safe to remove.
After the audio provider rework, adjust the audio players to not
use the int16 mono downmixed audio unless necessary. Furthermore,
the pulseaudio-based player now controls the volume directly through
pulseaudio instead of by modifying the buffer. This also reduces latency
when changing the volume.
The entire set of GetAudio functions is quite messy now. After wangqr's
audio rework, it was split into GetAudio and GetInt16MonoAudio
functions, but now volume scaling is also necessary. Really, this should
go back to a type constructor based system with audio players being
allowed to choose what properties out of mono / 16 bytes / int samples /
volume they need.
Checking if the diff is '>' instead of '>=' will always fail if
tolerance=0, even if the colours are identical. This will cause the line
to get a startime greter than its end time, which is not desireable.
Rounding errors and limits of floating type precision might still affect
the comparison. An additional sanity check before calculation is added
to ensure the selected position and colour match within tolerance.
This allows us to refactor the search code to never check the starting
frame and guanrantees valid timings with start<end even with
rounding errors.
No other tooltips end with points, even if they're a sentence.These
ones aren't even sentences.
Also updates all *.po files accordingly, with an hacked in
POT-Creation-Date due to issues with make_pot.sh.
Float edits with a spinner are by default sized to fit their full
min-max range of possible values, so this makes min and max default to 0
and 100 (like it'd done internally in wx) instead of -DOUBLE_MAX and
DOUBLE_MAX.
Note that this does change the behavior of lua dialogs, but does not
contradict existing documentation or specification. It should only
affect scripts who either disobey the specification by specifying only
one value out of max/min, or scripts displaying these large float edits
by specifing a step, but no max or min.
Starting with AVISYNTH_INTERFACE_VERSION=5, this is how script
environments should be deleted. The previous code was causing crashes
when unloading AviSynth in certain scenarios, such as when failing to
open a file due to an incorrect path.
- DS2 Player has a similar structure to XAudio, so I don't see any reason why not to enable 1 channel+ audio.
- Haven't tried with 5.1 channel sources but I believe it should be the same as 2 channel (As in, Directsound will downmix the audio to 2 channel)
- Moved the volume setting to using the player directly and from some quick audio tests, -10000 is too soft. I tried with -5000 instead which seems to be alright.
- To allow for XAudio2 to work properly, we need to rework how does provider work since they only are used to be able to take in mono audio.
- Other providers have been dumbed down to accept single channel audio since originally aegisub only accepted 1 channel audio.
- meson.build has been modified to accommodate for xaudio, as we currently don't accept redistributable forms of xaudio, we need to work around the WinNT version.
- There has been 1 more fix to res.rc to allow for compiling on non tagged releases.
Checking if the diff is '>' instead of '>=' will always fail if
tolerance=0, even if the colours are identical. This will cause the line
to get a startime greter than its end time, which is not desireable.
Rounding errors and limits of floating type precision might still affect
the comparison. An additional sanity check before calculation is added
to ensure the selected position and colour match within tolerance.
This allows us to refactor the search code to never check the starting
frame and guanrantees valid timings with start<end even with
rounding errors.
No other tooltips end with points, even if they're a sentence.These
ones aren't even sentences.
Also updates all *.po files accordingly, with an hacked in
POT-Creation-Date due to issues with make_pot.sh.
