Re: ghostscript fonts (was: new smaller installers to test)
On Fri, Feb 19, 2010 at 11:30 PM, Graham Percival gra...@percival-music.ca wrote: On Fri, Feb 19, 2010 at 8:44 PM, Han-Wen Nienhuys hanw...@gmail.com wrote: I am a little worried that it might depend on how the font is laid out internally. Perhaps we should let this slip for now, but look into this as the first suspect when problems with international font rendering pop up. We had one report on the bug list about fonts on windows. I've added the Fonts dir back. 2.13.13-2 is here: http://lilypond.org/~graham/ I can only test linux-x86 right now, and the test file Martial sent to bug-lilypond (invalid-font.ly) compiles fine with 2.13.13-2. Other than that, I agree that we should leave it for now. I have no problem adding some (all?) of these dirs back, but let's wait for more bug reports or other evidence that we actually need them. I agree. Thanks, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 19, 2010 at 5:50 AM, Graham Percival gra...@percival-music.ca wrote: On Thu, Feb 18, 2010 at 11:25:33PM -0200, Han-Wen Nienhuys wrote: The largest ones are the Encoding files, which are related to international fonts, so if this is the cause of the problem (I think it is). ok, I'll test that when I get to university in a few hours. Well, I tried the linux-x86 installer from yesterday's git, and it built utf-8.ly just fine. Do the binaries in GUB use the system-installer ghostscript setup if it can't find things from within GUB? I think it should not, but if unsure, you can remove the Resources dir from the system GS too. (Reinstall from OS source afterwards) It might be a good idea to drop a note on a GS mailing list to verify what functionality you'd loose. -- Han-Wen Nienhuys - han...@xs4all.nl - http://www.xs4all.nl/~hanwen ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Thu, Feb 18, 2010 at 11:50 PM, Graham Percival gra...@percival-music.ca wrote: Could other people try compiling a file with Chinese or Japanese symbols using the official 2.13.13 ? I'd like to know whether it's just the regtest-building that's broken, or the entire installer. The regtest utf-8.ly is compiling just fine (assuming a Japanese font is installed) with 2.13.13 linux-x86 and darwin-x86. Thanks, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 19, 2010 at 6:26 PM, Patrick McCarty pnor...@gmail.com wrote: Could other people try compiling a file with Chinese or Japanese symbols using the official 2.13.13 ? I'd like to know whether it's just the regtest-building that's broken, or the entire installer. The regtest utf-8.ly is compiling just fine (assuming a Japanese font is installed) with 2.13.13 linux-x86 and darwin-x86. I am a little worried that it might depend on how the font is laid out internally. Perhaps we should let this slip for now, but look into this as the first suspect when problems with international font rendering pop up. -- Han-Wen Nienhuys - han...@xs4all.nl - http://www.xs4all.nl/~hanwen ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
ghostscript fonts (was: new smaller installers to test)
On Fri, Feb 19, 2010 at 8:44 PM, Han-Wen Nienhuys hanw...@gmail.com wrote: On Fri, Feb 19, 2010 at 6:26 PM, Patrick McCarty pnor...@gmail.com wrote: Could other people try compiling a file with Chinese or Japanese symbols using the official 2.13.13 ? I'd like to know whether it's just the regtest-building that's broken, or the entire installer. Oops, I'd forgotten about: http://code.google.com/p/lilypond/issues/detail?id=934 GUB is missing Japanese fonts, and has been for months. This isn't related to the ghostscript issues at all. I am a little worried that it might depend on how the font is laid out internally. Perhaps we should let this slip for now, but look into this as the first suspect when problems with international font rendering pop up. We had one report on the bug list about fonts on windows. I've added the Fonts dir back. 2.13.13-2 is here: http://lilypond.org/~graham/ Other than that, I agree that we should leave it for now. I have no problem adding some (all?) of these dirs back, but let's wait for more bug reports or other evidence that we actually need them. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Thu, Feb 18, 2010 at 2:13 AM, Han-Wen Nienhuys hanw...@gmail.com wrote: On Fri, Feb 12, 2010 at 4:57 PM, Graham Percival gra...@percival-music.ca wrote: I've tweaked the list of dirs to remove from share/ghostscript/Resources. The resulting files are (on average) 5 megs smaller. linux-x86 works here for me. It looks as if this directory contains various character encoding related stuff. Can you run this through a couple of CJK files to see if that keeps working? Umm... maybe? A quick google search suggests that CJK are Asian fonts? ... woah, this is weird. The 2.13.12 regtest comparison shows them just fine: http://lilypond.org/test/v2.13.12-1/compare-v2.13.11-1/index.html (you can see the hiragana in utf-8.ly) the 2.13.13 regest doesn't show them changing: http://lilypond.org/test/v2.13.13-1/compare-v2.13.12-1/index.html however, they're missing from the 2.13.13 input/regression: http://lilypond.org/doc/v2.13/input/regression/df/lily-5e79810b.png I don't have the uploads/ files in my gub dir any more, so I can't track it down further. But I'm really surprised about the regtest comparison not showing anything! Anybody have suggestions? I could simply revert the change, but it seemed that people liked the idea of saving some space. Does anybody know ghostscript and want to suggest which of these dirs I should try keeping? +'share/ghostscript/*/Resource/{CMap,ColorSpace,Decoding}/', +'share/ghostscript/*/Resource/{Encoding,Font,SubstCID}/', I don't know offhand how large each of those dirs are, but if it would help, I could build the package with the stripping commented out, and tell you exactly which files there are, and where they are. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Thu, Feb 18, 2010 at 4:43 PM, Graham Percival gra...@percival-music.ca wrote: woah, this is weird. The 2.13.12 regtest comparison shows them just fine: http://lilypond.org/test/v2.13.12-1/compare-v2.13.11-1/index.html (you can see the hiragana in utf-8.ly) the 2.13.13 regest doesn't show them changing: http://lilypond.org/test/v2.13.13-1/compare-v2.13.12-1/index.html however, they're missing from the 2.13.13 input/regression: http://lilypond.org/doc/v2.13/input/regression/df/lily-5e79810b.png The regtest comparison uses the bounding boxes inside lilypond. Errors in ghostscript dont generate differences inside lilypond, so they are not registered in lilypond. Anybody have suggestions? I could simply revert the change, but it seemed that people liked the idea of saving some space. Does anybody know ghostscript and want to suggest which of these dirs I should try keeping? + 'share/ghostscript/*/Resource/{CMap,ColorSpace,Decoding}/', + 'share/ghostscript/*/Resource/{Encoding,Font,SubstCID}/', I don't know offhand how large each of those dirs are, but if it would help, I could build the package with the stripping commented out, and tell you exactly which files there are, and where they are. The largest ones are the Encoding files, which are related to international fonts, so if this is the cause of the problem (I think it is). You could also run the GS command by hand with -dQUIET switched off, to see if it complains of missing Resource files. -- Han-Wen Nienhuys - han...@xs4all.nl - http://www.xs4all.nl/~hanwen ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Thu, Feb 18, 2010 at 11:25:33PM -0200, Han-Wen Nienhuys wrote: On Thu, Feb 18, 2010 at 4:43 PM, Graham Percival gra...@percival-music.ca wrote: woah, this is weird. The 2.13.12 regtest comparison shows them just fine: The regtest comparison uses the bounding boxes inside lilypond. Errors in ghostscript dont generate differences inside lilypond, so they are not registered in lilypond. Good to know! I'll add a note to the CG. Anybody have suggestions? I could simply revert the change, but it seemed that people liked the idea of saving some space. Does anybody know ghostscript and want to suggest which of these dirs I should try keeping? + 'share/ghostscript/*/Resource/{CMap,ColorSpace,Decoding}/', + 'share/ghostscript/*/Resource/{Encoding,Font,SubstCID}/', The largest ones are the Encoding files, which are related to international fonts, so if this is the cause of the problem (I think it is). ok, I'll test that when I get to university in a few hours. You could also run the GS command by hand with -dQUIET switched off, to see if it complains of missing Resource files. Patrick says it complains, but it did this before the extra stripping. http://lists.gnu.org/archive/html/lilypond-devel/2010-02/msg00140.html (I'm not certain if GenericResourceDir is the warning you mean, though) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 19, 2010 at 4:51 AM, Graham Percival gra...@percival-music.ca wrote: On Thu, Feb 18, 2010 at 11:25:33PM -0200, Han-Wen Nienhuys wrote: The largest ones are the Encoding files, which are related to international fonts, so if this is the cause of the problem (I think it is). ok, I'll test that when I get to university in a few hours. Well, I tried the linux-x86 installer from yesterday's git, and it built utf-8.ly just fine. Do the binaries in GUB use the system-installer ghostscript setup if it can't find things from within GUB? Could other people try compiling a file with Chinese or Japanese symbols using the official 2.13.13 ? I'd like to know whether it's just the regtest-building that's broken, or the entire installer. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 12, 2010 at 4:57 PM, Graham Percival gra...@percival-music.ca wrote: I've tweaked the list of dirs to remove from share/ghostscript/Resources. The resulting files are (on average) 5 megs smaller. linux-x86 works here for me. It looks as if this directory contains various character encoding related stuff. Can you run this through a couple of CJK files to see if that keeps working? -- Han-Wen Nienhuys - han...@xs4all.nl - http://www.xs4all.nl/~hanwen ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 12, 2010 at 01:52:01PM -0800, Patrick McCarty wrote: *** Warning: GenericResourceDir doesn't point to a valid resource directory. the -sGenericResourceDir=... option can be used to set this. WARNING: /Unicode /Decoding resource is not accessible but it is useful for generating ToUnicode CMap. Hmm. I'm specifically stripping the /Decoding dir, so that makes sense. I don't see any /Unicode dir, but I _do_ see a /CMap dir (which also gets stripped). *shrug* If somebody reports a new problem with 2.13.13 or later, we could look at keeping the /Unicode and /CMap dirs (and also setting up that GenericResourceDir thing). Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
new smaller installers to test
I've tweaked the list of dirs to remove from share/ghostscript/Resources. The resulting files are (on average) 5 megs smaller. linux-x86 works here for me. Could we get a few tests for various OSes? http://lilypond.org/~graham/ (mingw is mingw-new.exe, to avoid a clash with the nsis 2.4.6 binary test. The mingw-new.exe also uses 2.4.6, so if that file works for you, there's no need to test the older lilypond-blah.exe) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 12, 2010 at 7:06 PM, Kieren MacMillan kieren_macmil...@sympatico.ca wrote: Hi Graham, Could we get a few tests for various OSes? http://lilypond.org/~graham/ Mac OS X 10.6 is aok. It also seemed that first compile was almost instantaneous, as compared with earlier upgrades -- is this a change, or am I imagining things? I would be very surprised if that was a real change, but then again, I've been surprised at cross-platform issues in the past. :) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
Hi Graham, Could we get a few tests for various OSes? http://lilypond.org/~graham/ Mac OS X 10.6 is aok. It also seemed that first compile was almost instantaneous, as compared with earlier upgrades -- is this a change, or am I imagining things? Cheers, Kieren. ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
Graham Percival wrote Friday, February 12, 2010 6:57 PM I've tweaked the list of dirs to remove from share/ghostscript/Resources. The resulting files are (on average) 5 megs smaller. linux-x86 works here for me. Could we get a few tests for various OSes? On Vista: Yes - the .exe is 20.5 Mb, previous versions have been c. 27 Mb. Size of installation on disk is now 68 Mb, previous versions around 83 Mb. Big improvement. Installation appeared perfectly normal. Ran lilypond-book and texi2html on pitches.itely. Output looks fine. First run took c. 2 mins to get going, second started immediately. Presumably fonts are being processed first time ?? Uninstall took exactly 10 mins, and left three files undeleted: /usr/bin/lilypad-unicode.exe /usr/share/lilypond/current/dvips/ps/lilyponddefs.ps /usr/share/lilypond/current/dvips/ps/music-drawing-routines.ps I don't know if this is a change in behaviour as I normally just delete or rename the lilypond directory. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: new smaller installers to test
On Fri, Feb 12, 2010 at 10:57 AM, Graham Percival gra...@percival-music.ca wrote: I've tweaked the list of dirs to remove from share/ghostscript/Resources. The resulting files are (on average) 5 megs smaller. linux-x86 works here for me. Could we get a few tests for various OSes? http://lilypond.org/~graham/ darwin-x86 and darwin-ppc work fine on clean 10.5 Leopard machines. I've noticed some warnings from Ghostscript (when using --verbose) in the past, and those are still present, so here they are for completeness: *** Warning: GenericResourceDir doesn't point to a valid resource directory. the -sGenericResourceDir=... option can be used to set this. WARNING: /Unicode /Decoding resource is not accessible but it is useful for generating ToUnicode CMap. ...though they don't seem to affect the output, so I guess they are harmless. Thanks, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel