Re: [PATCH 5/5] usp10/tests: Add shaping tests for fLogicalOrder and fRTL.
On 02.03.2017 09:38, Henri Verbeet wrote:
Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> --- dlls/usp10/tests/usp10.c | 153 +++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 148 insertions(+), 5 deletions(-)
I'm getting a lot of test failures after this series (starting with patch 2): $ (cd dlls/usp10/tests/; make test) ../../../tools/runtest -q -P wine -T ../../.. -M usp10.dll -p usp10_test.exe.so usp10 && touch usp10.ok fixme:uniscribe:GPOS_convert_design_units_to_device Font with lfWidth set not handled properly fixme:uniscribe:GPOS_convert_design_units_to_device Font with lfWidth set not handled properly usp10.c:1662: Test failed: 0: invalid LogClust(8) usp10.c:1662: Test failed: 1: invalid LogClust(7) usp10.c:1662: Test failed: 2: invalid LogClust(6) usp10.c:1662: Test failed: 3: invalid LogClust(6) usp10.c:1662: Test failed: 4: invalid LogClust(4) usp10.c:1662: Test failed: 5: invalid LogClust(3) usp10.c:1662: Test failed: 6: invalid LogClust(1) usp10.c:1662: Test failed: got incorrect number of glyphs (9) usp10.c:1662: Test failed: 2: uJustification incorrect (2) usp10.c:1662: Test failed: 2: fClusterStart incorrect (0) usp10.c:1662: Test failed: 3: fClusterStart incorrect (1) usp10.c:1662: Test failed: 3: fDiacritic incorrect (0) usp10.c:1662: Test failed: 3: fZeroWidth incorrect (0) usp10.c:1662: Test failed: 5: fClusterStart incorrect (0) usp10.c:1662: Test failed: 5: fDiacritic incorrect (1) usp10.c:1662: Test failed: 5: fZeroWidth incorrect (1) usp10.c:1662: Test failed: 7: uJustification incorrect (0) usp10.c:1662: Test failed: 8: uJustification incorrect (0) usp10.c:1662: Test failed: 8: fClusterStart incorrect (1) usp10.c:1662: Test failed: 8: fDiacritic incorrect (0) usp10.c:1662: Test failed: Got unexpected glyphProp2[6].sva.uJustification 0, expected 0x2. usp10.c:1662: Test failed: Got unexpected glyphProp2[6].sva.fClusterStart 0x1, expected 0. usp10.c:1662: Test failed: Got unexpected glyphProp2[7].sva.uJustification 0x2, expected 0. usp10.c:1662: Test failed: Got unexpected glyphProp2[7].sva.fClusterStart 0, expected 0x1. fixme:uniscribe:GPOS_convert_design_units_to_device Font with lfWidth set not handled properly fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:get_cluster_advance multi-glyph reversed clusters found fixme:uniscribe:GPOS_convert_design_units_to_device Font with lfWidth set not handled properly fixme:uniscribe:GPOS_convert_design_units_to_device Font with lfWidth set not handled properly fixme:uniscribe:SHAPE_GetFontFeatureTags Filtering not implemented Is this a problem with the fonts shipped on Archlinux, or is something going wrong on the Wide side? The font "Estrangelo Edessa" is provided by /usr/share/fonts/OTF/SyrCOMEdessa.otf which can be downloaded at https://www.archlinux.org/packages/extra/any/xorg-fonts-misc/download/. If you need any logs or want me to open a bug report, please let me know. Best regards, Sebastian
On 7 March 2017 at 04:30, Sebastian Lackner <sebastian(a)fds-team.de> wrote:
Is this a problem with the fonts shipped on Archlinux, or is something going wrong on the Wide side? The font "Estrangelo Edessa" is provided by /usr/share/fonts/OTF/SyrCOMEdessa.otf which can be downloaded at https://www.archlinux.org/packages/extra/any/xorg-fonts-misc/download/. If you need any logs or want me to open a bug report, please let me know.
It seems likely that it's related to the font in some way, although without looking further into it I couldn't say whether that's because the font is e.g. missing a ligature or usp10 applying a feature incorrectly. Do you know whether it works correctly with the font from e.g. Windows 7?
On 7 March 2017 at 08:21, Henri Verbeet <hverbeet(a)gmail.com> wrote:
On 7 March 2017 at 04:30, Sebastian Lackner <sebastian(a)fds-team.de> wrote:
Is this a problem with the fonts shipped on Archlinux, or is something going wrong on the Wide side? The font "Estrangelo Edessa" is provided by /usr/share/fonts/OTF/SyrCOMEdessa.otf which can be downloaded at https://www.archlinux.org/packages/extra/any/xorg-fonts-misc/download/. If you need any logs or want me to open a bug report, please let me know.
It seems likely that it's related to the font in some way, although without looking further into it I couldn't say whether that's because the font is e.g. missing a ligature or usp10 applying a feature incorrectly. Do you know whether it works correctly with the font from e.g. Windows 7? So I looked into this a bit, and it looks like that while the Microsoft font has a single glyph for the lamadh-alaph ligature, the Arch font constructs that ligature out of 3 separate glyphs. Both are legitimate, but they're different from each other.
On 07.03.2017 14:23, Henri Verbeet wrote:
On 7 March 2017 at 08:21, Henri Verbeet <hverbeet(a)gmail.com> wrote:
On 7 March 2017 at 04:30, Sebastian Lackner <sebastian(a)fds-team.de> wrote:
Is this a problem with the fonts shipped on Archlinux, or is something going wrong on the Wide side? The font "Estrangelo Edessa" is provided by /usr/share/fonts/OTF/SyrCOMEdessa.otf which can be downloaded at https://www.archlinux.org/packages/extra/any/xorg-fonts-misc/download/. If you need any logs or want me to open a bug report, please let me know.
It seems likely that it's related to the font in some way, although without looking further into it I couldn't say whether that's because the font is e.g. missing a ligature or usp10 applying a feature incorrectly. Do you know whether it works correctly with the font from e.g. Windows 7? So I looked into this a bit, and it looks like that while the Microsoft font has a single glyph for the lamadh-alaph ligature, the Arch font constructs that ligature out of 3 separate glyphs. Both are legitimate, but they're different from each other.
I just tested it, and everything seems to work with the original estre.ttf from Windows 7 (after moving the font in /usr/share). What is your suggestion, should we accept the Archlinux version as valid, or maybe skip the test if this replacement is installed? Best regards, Sebastian
On 7 March 2017 at 14:42, Sebastian Lackner <sebastian(a)fds-team.de> wrote:
I just tested it, and everything seems to work with the original estre.ttf from Windows 7 (after moving the font in /usr/share). What is your suggestion, should we accept the Archlinux version as valid, or maybe skip the test if this replacement is installed?
Probably just skip, although I wouldn't be opposed to accepting multiple results either. In the longer term, I think ideally we'd include our own font for these tests, but it may take some effort to create one that covers all the different scripts.
participants (2)
-
Henri Verbeet -
Sebastian Lackner