Re: Building GNUstep on Solaris

[Fred Kiefer]

Am 19.01.2010 23:30, schrieb Andreas Höschler:
> route, so it seems that GNUstep really requires libffi now and no longer
> works with ffcall!

This is not true, we just recommend libffi, as it generally works better.

> I installed Affiche, TextEdit and Terminal. None of the apps work, all
> do core dump upon startup! :-(
> 
> gdb /opt/GNUstep/Local/Applications/Affiche.app/Affiche /core
> 
> Program terminated with signal 11, Segmentation fault.
> #0  0xd0f4b6ed in FTC_CMapCache_Lookup (cmap_cache=0x8187ff0,
> face_id=0x81ef558, cmap_index=0, char_code=64256) at
> /usr/share/src/freetype-2.1.10/src/cache/ftccmap.c:266
> 266         gindex = node->indices[char_code - node->first];
> 
> #0  0xd0f4b6ed in FTC_CMapCache_Lookup (cmap_cache=0x8187ff0,
> face_id=0x81ef558, cmap_index=0, char_code=64256) at
> /usr/share/src/freetype-2.1.10/src/cache/ftccmap.c:266
> #1  0xd1035ea5 in -[FTFontInfo glyphForCharacter:] (self=0x82222d8,
> _cmd=0xd1f67770, ch=64256) at ftfont.m:2253
> #2  0xd1d1c85d in -[NSGlyphGenerator
> generateGlyphsForGlyphStorage:desiredNumberOfCharacters:glyphIndex:characterIndex:]
> (self=0x80a7810, _cmd=0xd1f9eb60,
>     storage=0x825f310, num=7, glyph=0x80452a8, index=0x80452ac) at
> NSGlyphGenerator.m:186
> #3  0xd1e6edbc in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=0, h=0x825ce08, stop=0x0,
>     all_complete=0x8045333 "\001") at GSLayoutManager.m:686
> #4  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=1, h=0x825f470, stop=0x0,
>     all_complete=0x80453b3 "\001") at GSLayoutManager.m:667
> #5  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=2, h=0x825f460, stop=0x0,
>     all_complete=0x8045433 "\001") at GSLayoutManager.m:667
> #6  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=3, h=0x825f450, stop=0x0,
>     all_complete=0x80454b3 "\001") at GSLayoutManager.m:667
> #7  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=4, h=0x825f440, stop=0x0,
>     all_complete=0x8045533 "\001") at GSLayoutManager.m:667
> #8  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=5, h=0x825f430, stop=0x0,
>     all_complete=0x80455b3 "\001") at GSLayoutManager.m:667
> #9  0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=6, h=0x825f420, stop=0x0,
>     all_complete=0x8045633 "\001") at GSLayoutManager.m:667
> #10 0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=7, h=0x825f410, stop=0x0,
>     all_complete=0x80456b3 "\001") at GSLayoutManager.m:667
> #11 0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=8, h=0x825f400, stop=0x0,
>     all_complete=0x8045733 "\001") at GSLayoutManager.m:667
> #12 0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=9, h=0x825f3f0, stop=0x0,
>     all_complete=0x80457b3 "\001") at GSLayoutManager.m:667
> #13 0xd1e6ed24 in -[GSLayoutManager(GlyphsHelpers)
> _generateGlyphs:char:r:::::::] (self=0x825f310, _cmd=0xd1f9eb58, last=0,
> cpos=0, gpos=0, level=10, h=0x825f3e0, stop=0x0,
>     all_complete=0x8045833 "\001") at GSLayoutManager.m:667
> ...

This is not a loop in GNUstep, as your other mail suggests. The
recursion is working as expected as you can see on the decreasing level
parameter (Don't expect me to explain the details of the GSLayoutManager
here, you really don't want to know).
What goes wrong is the glyph lookup in the font. The character it
searches the glyph for is 64256. This is the "ff" ligature and it your
font doesn't support it, it should just return NULL from the function
FTC_CMapCache_Lookup. But there is one thing that could be strange here,
the cmap_index is 0, this could be correct or the result of a missing
FT_ENCODING_UNICODE encoding in this font. The code in [FTFontInfo
-initWithFontName:matrix:screenFont:] doesn't differentiate between
these two cases.
Looks like your font is not suitable for GNUstep and the code doesn't
detect this correctly.

Fred