Commit graph

486 commits

Author SHA1 Message Date
Carl Worth
ccf60202d3 Add SVG-specific reference image for source-clip-scale test
The difference here is that the SVG backend uses a meta-surface for
its similar surface, so there are no rasterization/filtering
artefacts when scaling the source surface.
2006-08-17 13:35:16 -07:00
Carl Worth
64d2feb9f6 Harmonize implementations of source-clip and source-clip-scale to make similarities more evident. 2006-08-17 13:35:15 -07:00
Carl Worth
524507c39f Rename src-clip test to source-clip-scale 2006-08-17 13:35:15 -07:00
Vladimir Vukicevic
afb50580ce Add src-clip test case to demonstrate bug with clipping applying to a source surface. 2006-08-17 13:35:15 -07:00
Behdad Esfahbod
6122cc85c8 [test] Add $(LDADD) to svg2png_LDADD, to link to correct cairo
Previously it was using the cairo found in system.
2006-08-16 20:07:06 -04:00
Behdad Esfahbod
30cd66fe9a [test] Add ft-text-vertical-layout-truetype. 2006-08-16 19:43:55 -04:00
Behdad Esfahbod
158e160211 [.gitignore] Add truetype-tables 2006-08-16 19:43:41 -04:00
Behdad Esfahbod
cbcf1ddd64 [test] New torture tests ft-text-vertical-layout-truetype/type1
that combine vertical layout with font_matrix rotation and translation.
Currently Type3 and Type1 font paths both show broken behavior.
The type1 test uses "Nimbus Sans L" as the font.
2006-08-16 14:27:40 -04:00
Behdad Esfahbod
91eb56998c [test] ft-text-vertical-layout: Update ref images 2006-08-15 05:49:21 -04:00
Behdad Esfahbod
db779b5121 [test] ft-text-vertical-layout: Use the string "i-W" to better show vertical metrics 2006-08-15 05:27:59 -04:00
Emmanuel Pacaud
17ec33aa4f New test for long line bug.
This shows at least an issue in cairo_fixed_from_double where values
> 32767 are converted to -32768, instead of being clamped to 32767.
2006-08-14 20:19:15 +02:00
Pavel Roskin
a06af40c35 [test] Fix warnings in tests on 64-bit systems
Cast argiments from size_t to int.  size_t is 64-bit on 64-bit systems,
which causes a warning.  The actual data should fit 32 bit comfortably.
2006-08-13 04:47:11 -04:00
Pavel Roskin
58963a2bf9 [test] Allow overriding Valgrind flags on the command line
Don't hardcode Valgrind flags in tests/Makefile.am so that
tests/Makefile doesn't need to be rebuilt to use different flags.  Not
everybody is looking for memory leaks.
2006-08-13 04:44:39 -04:00
Behdad Esfahbod
affc9c71ac [Makefile.am] Add $(AM_MAKEFLAGS) to recursive make invocations 2006-08-11 00:24:47 -04:00
Behdad Esfahbod
ee885c76af [test] Update EXTRA_DIST after beos_bitmap -> beos-bitmap change 2006-08-10 18:54:18 -04:00
Behdad Esfahbod
915448e50f [test] Add target check-ref-dups that lists reference images having duplicate sha1sum hash
This is useful to review every once in a while to drop duplicate images that
can be dropped.  For example, if there are two identical images one named
some-test-svg-rgb24-ref.png and other some-test-svg-argb32-ref.png, those two
can be replaced with some-test-svg-ref.png.
2006-08-10 15:35:36 -04:00
Behdad Esfahbod
cecf396f6f [test] Point out to the test log file on failures 2006-08-10 15:11:46 -04:00
Behdad Esfahbod
15074cbb04 [test] Behave better if no ref image was found for a test 2006-08-10 14:58:33 -04:00
Behdad Esfahbod
67ff765e30 [test] Rename beos_bitmap target to beos-bitmap. Same for directfb-bitmap. 2006-08-10 13:53:24 -04:00
Carl Worth
e3c7840d22 Fix EXTRA_DIST so that 'make distcheck' works again 2006-08-08 14:29:29 -07:00
Carl Worth
ca99478d43 Fix some memory leaks in a few of the tests. 2006-08-08 12:48:56 -07:00
Carl Worth
d6e204b9d5 Add a variation of an existing valgrind suppression 2006-08-08 12:48:20 -07:00
Carl Worth
cb5edb6c35 Eliminate warning due to the test suite's private 'FLATTENEND' format value 2006-08-08 01:39:12 -07:00
Behdad Esfahbod
f089a07679 Fix misplaced volatile keyword 2006-08-08 04:36:13 -04:00
Carl Worth
c3b912d7db Eliminate most compiler warnings from the test suite 2006-08-08 01:16:49 -07:00
Carl Worth
c7d11ad2b4 Remove stale SVG-specific reference images for pixman-rotate test 2006-08-08 00:43:09 -07:00
Carl Worth
401f0ce3c4 Merge branch 'surface-font-options' into cairo 2006-08-08 00:19:51 -07:00
Behdad Esfahbod
9136c21770 Update reference images, mostly PDF and SVG
Also replace some SVG ref images for rgb24 and argb32 with a single
one where the two have been similar.
2006-08-07 23:47:31 -07:00
Behdad Esfahbod
56791ab31d Make "make html" not trigger test reruns
"make index.html" can still be used to force up-to-date test results,
causing test reruns if necessary.
2006-08-08 02:46:38 -04:00
Behdad Esfahbod
c2ea56e5e0 Add test truetype-tables that checks sizeof truetype tables 2006-08-07 21:37:16 -04:00
Behdad Esfahbod
127704c225 Look for per-target reference image too 2006-08-01 20:39:56 -04:00
Behdad Esfahbod
fc715ffd80 New PS ref images, matching the recent png16m->pngalpha driver change 2006-08-01 18:21:27 -04:00
Behdad Esfahbod
3e24f516a2 Change the text string "cd" to "AB" in test to match the other string
that is drawn.  The reason, one is draw using show_text, the other
using text_path.  The difference is more emphasized when rendering the
same string.
2006-08-01 16:13:29 -04:00
Behdad Esfahbod
3d95919fab Switch from ghostscript's png16m driver to pngalpha for PS->PNG again.
We have switched back and forth quite a few times.  This time I'm switching
because with pngalpha we get gray antialiased text and graphics while with
png16m all we get is no antialiasing.  This is definitely a bug in the png16m
driver, but I won't wait until it gets fixed upstream.

Previously Carl Worth switched to pngalpha and reverted it immediately in
commit c4fc7b06b5.  I've now fixed image-diff to
work with the output of pngalpha, so we can switch.  It requires lots of
reference image updates, but still doesn't help with reducing the number of
PS-specific reference images we need.
2006-08-01 15:27:11 -04:00
Behdad Esfahbod
5a23fd70a0 Change the way diff images highlight differences.
Previously it was using the equation 128+diff/3, which results in
lots of gray and de-emphasized difference.  Now it's using
MIN(255,diff*4) which more emphasizes the real difference.
2006-08-01 15:20:39 -04:00
Behdad Esfahbod
d85f30e789 Make image_diff_flattened flatten the first image too.
This is useful to use the ghostscript pngalpha driver for example.
2006-08-01 15:18:19 -04:00
Behdad Esfahbod
556a4d8405 Add vector_ignored_tests that is tests ignored for ps/pdf/svg
that includes all tests depending on CAIRO_ANTIALIAS_NONE and
CAIRO_ANTIALIAS_SUBPIXEL.
This removes separate pdf_ignored_tests and svg_ignored_tests
arrays that were out of synch and otherwise the same.
2006-08-01 15:16:10 -04:00
Behdad Esfahbod
2fa709d6af Set font options in the test context and make tests not do that
This should help with not requiring many backend-specific reference
images, and some should be removed now.
2006-07-31 15:17:15 -04:00
Behdad Esfahbod
e92d693e07 Make "CAIRO_TEST_TARGET=... make check" work again
Do a funny transition of CAIRO_TEST_TARGET through TARGETS such that
one can limit tested targets both through CAIRO_TEST_TARGET env var
and TARGETS make var on the command line.
2006-07-27 14:59:53 -04:00
Behdad Esfahbod
47d3c5a2c6 Flush stdout after writing \r, to remove artifacts. 2006-07-14 20:06:34 -04:00
Carl Worth
c70edff084 Add font-matrix-translation test case. 2006-07-14 14:40:47 -07:00
Carl Worth
b390e2c456 pixman-rotate: remove stale svg backend reference images 2006-07-14 14:40:47 -07:00
Carl Worth
2b30f85cad ft-text-vertical-layout: Update stale reference images 2006-07-14 14:40:47 -07:00
Behdad Esfahbod
6a5d66f651 Go back to "!!!CRASHED!!!" message now that there's no confusion. 2006-07-13 18:53:25 -04:00
Behdad Esfahbod
6c7099b0cc Only generate HTML for log files passed on the command line.
With the make bits already in place, this means that make retest
creates HTML for only the retested tests.
2006-07-13 18:50:14 -04:00
Carl Worth
4b22cb41a9 Use 'FAIL' instead of 'UNEXPECTED FAILURE' in test output.
There's no ambiguity anymore due to duplicated printing, so go
back to the classic, and easier to read wording.
2006-07-13 15:23:06 -07:00
Carl Worth
b2d04d7f1a Rename no_fail_on_stdout to eliminate confusing negative inside a Boolean variable name 2006-07-13 15:21:02 -07:00
Carl Worth
ff1280ce98 cairo-test.c: Prefer FALSE and TRUE for cairo_bool_t values. 2006-07-13 15:17:57 -07:00
Behdad Esfahbod
ebc7367306 Fix test result formatting. 2006-07-13 18:18:08 -04:00
Behdad Esfahbod
b2668e944d Use cairo_bool_t. 2006-07-13 14:02:58 -04:00