| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Now that we aren't using the M_ constants, drop math.h.
|
|
|
|
|
|
|
|
|
|
| |
Math implementations are platform dependent, and for
example msvc requires a define before enabling the
constants [1]
Use G_SQRT2 implementation which is independent instead.
[1] https://docs.microsoft.com/en-us/cpp/c-runtime-library/math-constants?view=msvc-160
|
|
|
|
|
| |
The api is deprecated, but it is very ugly to randomly
miss a few values here. Just add them.
|
| |
|
|
|
|
| |
Exercise the list model apis, for coverage.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Check the previous change.
|
|
|
|
| |
These improve converage for misc PangoContext apis.
|
|
|
|
| |
These improve test coverage for some corner cases.
|
| |
|
|
|
|
| |
Cover many more error cases.
|
|
|
|
|
|
|
| |
Add a way to deserialize a PangoAttrList from a string,
for less cumbersome tests.
And use it in testattributes.
|
|
|
|
|
| |
This tests that we handle more cases of overlapping
attributes properly.
|
|
|
|
| |
Test the fix in the previous commit.
|
|
|
|
| |
This captures bug fixed in the previous commit.
|
|
|
|
|
|
|
|
|
| |
The commit message for d476d2a944500f8553 said
"word boundaries", but the code checked for line
breaks. That was wrong, since it also made hyphens
disappear at SHY, where we very much want them.
Update affected tests.
|
|
|
|
|
|
|
| |
test-break has a helpful text that explains
the maining of the output. Unfortunately,
the --help option is taken by g_test_init(),
so make this available with --legend.
|
|
|
|
|
|
|
| |
We were just never setting the height. Ouch!
Test included.
Fixes: #487
|
|
|
|
|
|
|
|
|
|
| |
We only want to insert hyphens when we are breaking
inside a word.
Update affected tests (hyphen-or-not shows up in the
analysis flags in the output of test-layout).
Fixes: #558
|
|
|
|
|
|
| |
When using test-layout FILE to generate
expected output, we don't want to prepend
an extraneous '#' to every line.
|
|\
| |
| |
| |
| | |
Markup parse fixes
See merge request GNOME/pango!292
|
| |
| |
| |
| |
| | |
Add tests that test some of the corner cases handled
in the previous commit.
|
| |
| |
| |
| |
| | |
Don't print extraneous output when we are generating
an .expected file.
|
| |
| |
| |
| |
| |
| | |
And print out the accel_char if we get one back.
This will let us add tests for accel functionality.
|
| |
| |
| |
| |
| | |
This is in preparation for adding some tests that
will check accel handling.
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the 'other' list contains attributes that are
unlimited or exceed the range given to pango_attr_list_splice,
those attributes were 'leaking' out of the range. The visible
effect of this is that the underline of preedit text extends
outside the preedit in some GTK entries.
Fix this by clipping the inserted attributes to the range.
The documentation is not very explicit about this, but I believe
this is the expected behavior.
Tests included.
|
|
|
|
|
| |
Remove the dashes that cause the runs to break, and seem
to be causing issues on both Debian and newer Fedora.
|
|
|
|
|
|
|
|
|
| |
Test that round-tripping through pango_font_describe
works. This is currently broken for scalable bitmap
fonts, such as color Emoji fonts.
We skip the test on OS X where we are most likely
missing the fonts.
|
|
|
|
| |
pango_cairo_font_map_get_default is transfer none.
|
|
|
|
|
|
|
| |
When producing output, use the C locale for formatting
floating point numbers, to avoid locale dependent output.
Fixes: #474
|
|\
| |
| |
| |
| | |
Fix test leak
See merge request GNOME/pango!276
|
| |
| |
| |
| |
| | |
We just plugged all the memory leaks in tests,
lets not introduce new ones.
|
|/
|
|
|
|
|
|
|
|
| |
We can't guarantee uniqueness of face names, so we
don't know that getting a face by name will return
the same face we got the name from (which is true if
faces are unique). So, just assert that we get a
face with the same name.
Fixes: #494
|
|\
| |
| |
| |
| |
| |
| | |
Fix the length checking
Closes #526
See merge request GNOME/pango!273
|
| |
| |
| |
| | |
Closes #526
|
| | |
|
|/ |
|
|
|
|
|
|
|
|
|
|
|
| |
When versions of Meson with bug
https://github.com/mesonbuild/meson/issues/7515 are no longer widespread,
the Pango test suite should diagnose which tests are skipped again.
Versions 0.55.0 to 0.55.2 inclusive have the bug; version 0.55.3 is OK.
This reverts commit 440f9c2609b7f5f9d8f65619b029a5eb660de88b.
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
|
|
| |
This fixes the build after commit ddc282f6ca84dc6e6e5586a0addfa82a74e6e761
that changed how the scripts are invoked, but didn't add the executable
bits.
|
|
|
|
|
| |
test-shape was using one of its attribute lists
after dropping the reference on it. Don't do that.
|
|\
| |
| |
| |
| | |
Test shape fix
See merge request GNOME/pango!257
|
| |
| |
| |
| |
| |
| |
| | |
We don't have any test cases for this in our testsuite,
but it is useful to run this manually to see the shaping
results. The output was missing all but the first char
for the last cluster in an item.
|
|/
|
|
|
|
|
| |
As pointed out by Werner Lemberg, we were adding
four.break twice if libthai is used.
Fixes: #508
|
|
|
|
|
|
|
|
|
|
|
|
| |
We are using the deprecated python3 module for historical reasons. We
used to support both Python 2 and Python 3 in the Autotools build, but
since we switched to Meson we *know* we have Python 3 available. This
allows us to use a shebang line in the Python scripts we use in the
build, instead of invoking them through the Python interpreter in a
custom target; Meson knows how to handle shebangs portably as well.
This change removes the only deprecation warning coming from Meson when
configuring the Pango build.
|
|
|
|
|
| |
This tests that we are not breaking items up
in the middle of a keycap sequence.
|
|\
| |
| |
| |
| | |
misc fixes
See merge request GNOME/pango!242
|
| |
| |
| |
| | |
This was broken in f6032b754fd76ebbf0ae6.
|