Our handling of Photoshop's Blend Clipped Layers as Group is flawed
causing the order of layers to be messed up which may cause certain
parts of the image to become invisible or having the wrong colors.
This only fixes the incorrect layer ordering. Correct handling of
group layers with clipping needs a more extensive rewrite.
This is a manual revert of eac1df4b61
since other changes were made to the same lines.
Although using legacy layer modes for several blend modes improves
compatibility with Photoshop, we found out during testing of the
image from issue #10498, that it negatively impacts loading of PSD's
that use Blend Clipped Layers as Group.
We convert this specific PSD setting by adding a layer group and
setting Composite Mode to Blend to Background.
However, our legacy layer modes do not support Composite Mode, meaning
that PSD's using this can show up using smeared colors.
Since this hasn't been in a release yet, let's revert this and rethink
how we want to handle the combination of these PSD settings.
Resolves#9933.
We used png_get_valid (pp, info, PNG_INFO_tRNS) to check if
an indexed PNG had transparency when exporting.
However, this function does not return TRUE/FALSE but instead
the contents of that PNG chunk. Since we use index 0 for transparency,
this function always returns 0. The non-transparent indexed export code
ran regardless, creating incorrect PNGs.
We set pngg.has_trns earlier if the image has transparency, so it's safer
to use this variable instead.
Magic number 1 changed to 2 in one place. Missed in earlier fix for #10127.
Changed magic number to a defined macro constant so it might not happen again,
should GimpConfig change again.
Substantive change is call gimp_item_get_by_id instead of gimp_drawable...
And a few type changes from GimpDrawable* to GObject*.
Lots of renaming of functions, substituting "item" for "drawable."
Update comments.
Made one function static that was already private.
Fix the test case gimp_vectors_set_selected
Add more tests
Enhance testing framework, add function assert-PDB-true
Enhance some tests: make test files order independent
Fix tests broken by recent changes to GIMP
There was the question on whether it needed an hyphen or not. The hyphen
naming was often used to describe the "Hard-edge painting" style, though
even for this, it was not an absolute case. Depending on English writing
style, we could also find "hard edge painting". Also Liam was noting
that when the term appears by itself, as a noun (rather than a
qualificative), it was likely best without hyphen. So we went for the
no-hyphen wording.
Then there was the capitalization question: should capitalization rules
of data labels apply to parts between parentheses? I went with the
answer that *yes*, since it's part of the title anyway. And without the
part between parentheses, we'd have several data named the same. So it's
just as important.
As a consequence, I also re-capitalized "FG to BG (HSV
Counter-Clockwise)" and "FG to BG (HSV Clockwise Hue)" gradients,
following this updated rule.