When exporting a TIFF file without merging the image, make sure non
of the exported layers has a negative offset, by offsetting all the
layers as necessary. TIFF doesn't support negative page offsets,
giving an error in libtiff.
When loading a TIFF file with signed-int samples, convert the
samples to unsigned-int by offsetting them to the unsigned range,
instead of misinterpreting them as unsigned values.
The GDK_WINDOWING_X11 build-time macro check is not enough as GDK can be
built with both X11 and Wayland backends. We need to add a runtime check
of the type of display.
So it seems that pango_attr_iterator_range() could return G_MAXINT for
a Pango attribute when it is at the end of the preedit string. Looking
at Pango code, I see they initialize the attribute end property to
PANGO_ATTR_INDEX_TO_TEXT_END (G_MAXUINT), later clamped to G_MAXINT by
pango_attr_iterator_range(). So I guess for the specific case where we
are at the text end, it is normal. Only weird thing is that this didn't
happen at all on X11, only in Wayland.
So let's do our own pre-check. Also double the check by adding a UTF-8
validation.
This fixes preedit text not being displayed and the following warning:
> Gtk-CRITICAL **: 12:31:25.118: gtk_text_buffer_emit_insert: assertion 'g_utf8_validate (text, len, NULL)' failed
Even worse, this was potentially an out-of-range reading, though
fortunately checked early enough.
In gimp.css, don't set a minimum height for GimpDisplayShell
statusbars. Instead, in GimpStatusbar, set the widget's minimum
height to the maximum of its children's natural heights. Note that
we have to do this manually, instead of using a size group, since
GtkSizeGroup::ignore-hidden is deprecated (and nonfunctional) in
GTK3.
In begin_run() and end_run(), sync string properties set to
GIMP_ARGUMENT_SYNC_PARASITE with image parasites of the same name,
exactly the way "gimp-comment" was handled by begin_export() and
end_export(). Remove the "gimp-comment" handling code from
begin_export() and end_export().
which can be set to GIMP_ARGUMENT_SYNC_NONE (the default) or
GIMP_ARGUMENT_SYNC_PARASITE, which indicates that the argument should
be synced with an image parasite of the same name.
Align GimpSpinScale with gimp-2-10, by modifying its appearance and
behavior to match the 2.10 compact style, fixing interaction along
the way. Unlike 2.10, there is no option to revert to the old
style.
After discussion with Carlos Garnacho, we came to the conclusion this
list is a useless feature. Basically what we call "input device" here is
pretty much "pointer device" only. We are indeed explicitly ignoring any
device identified as GDK_SOURCE_KEYBOARD, leaving us only with various
types of pointer devices (and pads actually, which maybe we should also
ignore in fact).
Such devices don't usually come with "keys", only "buttons". And in rare
cases of very weird devices coming with both buttons and keys, they will
usually identify as 2 separate devices (a pointer device and a keyboard
one) anyway, in Carlos experience, so we would still wouldn't have
access to the real keys anyway.
Moreover these keys were not only useless, but also sometimes confusing,
because some pointer devices would actually list keys, but then if you
tried to map some key event, it would not do anything (as they are not
real keys). The tablets I was testing with were such, reporting hundreds
of keys which do nothing and only confused the hell out of me.
Carlos says it probably means that the tablet drivers send bogus data of
key descriptions (so a bug in the driver, harmless yet still confusing).
So let's just get rid of this key list as our tablet expert says these
are bogus and let's see if anyone ever reports feature loss of some
extra weird pointing device which one would have used in GIMP while
mapping keys. Note that I leave the concept of keys inside
GimpDeviceInfo for now (only removing the widget part listing these)
just in case we realize we have to bring these back (less chance of code
conflict in the future when reverting the small GUI commit). But chances
are high that we will have to clean GimpDeviceInfo too and just get rid
of key code there.
In other dialogs, it is not a revert to how it was before opening the
dialog, but a reset to default settings.
To just revert to dialog opening values, we can just use "Cancel" and
reopen the dialog (a bit cumbersome, but not something done often
anyway).
Currently what "Reset" does is to set back the device mode and any
customized axe curve. It doesn't touch customized keys, but these will
disappear anyway in a further commit.
Clean up faint pixels in the "Acrylic 05" brush, which produce a
badly-thresholded mask with the Pencil tool. See issue #5180.
Thanks to Sevenix for the fixed brush!
Fix horizontal downscaling of brush mipmap levels with odd width.
We'd previously fail to skip the last pixel of each input row,
which isn't included in the output when the width is odd, causing
subsequent output rows to be shifted to the right.
