Re: ANNOUNCE: Tk-804.027-p3.patch.gz
Slaven Rezic <[email protected]> 26 Nov 2006 17:24:38 +0100
| Newsgroups | gmane.comp.lang.perl.tk |
|---|---|
| Message-ID | <[email protected]> |
<[email protected]> writes: > Hi Slaven, > > thank you very much for the useful hint! Yes this was the major > problem: my libfontconfig.so did not know a single font! > I recompiled fontconfig-2.4.1 with some sensible paths to fonts, > and now your one-liner works, as well as several of the tests. > But many others still fail with segfault: > > Failed Test Stat Wstat Total Fail Failed List of Failed > ----------------------------------------------------------- > t/JP.t 0 139 294 0 0.00% ?? > t/KR.t 0 139 271 0 0.00% ?? > t/create.t 0 139 573 1046 182.55% 51-573 > t/entry.t 0 139 336 524 155.95% 75-336 > t/fbox.t 0 139 6 6 100.00% 4-6 > t/fileevent.t 0 139 1 0 0.00% ?? > t/font.t 0 139 14 2 14.29% 14 > t/geomgr.t 0 139 19 6 31.58% 17-19 > t/list.t 0 139 6 0 0.00% ?? > t/listbox.t 0 139 537 996 185.47% 40-537 > t/msgbox.t 0 139 13 20 153.85% 4-13 > t/text.t 0 139 119 0 0.00% ?? > t/unicode.t 0 139 7 4 57.14% 6-7 > t/wm-tcl.t 0 139 315 348 110.48% 142-315 > > I went back to fontconfig-2.3.2 and it worked again. > Same on Linux. On both Solaris and Linux, I can get a > working Tk with XFT with: > freetype-2.2.1 > fontconfig-2.3.2 > renderproto-0.9.2 > libXrender-0.9.2 > libXft-2.1.11 > But with the latest fontconfig-2.4.1 I get segfaults in > the above tests - and it seems that the segfault always > happens in some destruction function (FcPatternDestroy, > FcCharsetDestroy). Sorry, no idea whether this is Tks or > fontconfig's fault. I can reproduce the same (or a similar) error list on a debian linux machine with fontconfig 2.4.1, too. It is easily reproducible by creating a widget, destroying it (this is the important bit), and creating another widget. The problem vanishes if the FiniFont call in TkpDeleteFont in tkUnixXft.c is skipped. I also don't know yet if it's a Perl/Tk or fontconfig problem, but it smells like a problem of the latter. > > With the above configuration (the working one), I get > however still some errors in t/listbox.t (on Linux), see > attached log. Maybe also some issue with font sizing? Yes, determining when font-related tests should be skipped were wrong. Due to my (default FreeBSD?) fontconfig configuration, I get the original X11 font if helvetica was requested: $ fc-match helvetica helvR12-ISO8859-1.pcf.gz: "Helvetica" "Regular" The skips will be correct in the next version of the patch, hopefully. > > Hope this helps to make the Tk run with the latest and > greatest prerequisite packages... > > Cheers, > > Marek > Thanks, Slaven -- Slaven Rezic - slaven <at> rezic <dot> de tktimex - time recording tool http://sourceforge.net/projects/ptktools/ --++**==--++**==--++**==--++**==--++**==--++**==--++**== ptk mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/ptk