This is to remind us that we're motivated to make a new release of std postscript font set based on the the free URW fonts. * Some new glyphs have been submitted, but Filippov hasn't responded * The 8.11 version we released was built with a version of pfaedit that wrote a non-executable ([] vs {}) FontBBox array. While this is fine according to spec, some broken files expect it to be executable. Newer pfaedit versions correct this oversight, so re-exporting the fonts should take care of the issue. * The hints still need some review and cleanup. For example, the lowercase 'k' in n022003 still has overlapping hints.
I just want to add that most of new fonts doesn't work when installed into Windows 2000 (version shipped with gs till version 8.10 was ok)
Could you be a little more specific?
The additional glyphs for Uzbek are described here:
The additional glyphs for Uzbek are described here: http://www.ghostscript.com/pipermail/gs-devel/2003-December/002811.html my (much later) response is here: http://www.ghostscript.com/pipermail/gs-devel/2004-February/002936.html
Valek Filippov added my contributions and made a new tarball at ftp://ftp.gnome.ru/fonts/urw/release There is an issue with Nimbus Mono font. I reported a bug to Valek.
I would like to add, relative to Additional Comment #1, that the failure of the URW fonts to work in Windows 2000 is also found in Windows XP, and seems to have started last fall with SP4. According to some reports on usenet, MS quietly tightened up its standards for accepting what it considers "nonstandard" features in fonts, causing some fonts which had previously worked fine, not to any more. For me this is true of all the GS URW fonts (double-clicking on the PFM file used to bring up a fontview display; now it gives an error saying that this is not a valid font file). See MS Knowledge Base article 827487, which seems to confirm this. I have additionally seen this break happen to some other Type1 fonts, notably the BaKoMa fonts in the TeX archives, so I don't think it is limited to GS alone. Although I am almost totally ignorant of the inner details of fonts, I played around with the NOAH font editor, opening a URW font's PFB file, and then saving it and generating a new PFM file in the process. The new files seem to work just fine in Windows XP and 2000, for what it's worth. It would be good to fix the official GS fonts release to work properly in the latest patches of the Windows OS.
A follow-up to my comment #6 above: I just installed the latest version (8.14) of APFL Ghostscript, and this time I uninstalled all the fonts before installing the new version. All the Type1 GS fonts are now treated as valid by Windows xp and 2000. I notice that the file dates of all the fonts have gone back to older ones (2000) than the ones I had uninstalled.
Yes, it looks like Ray reverted to the older font release for gs814w32. It's hard to tell from your description exactly what the problem is. I did wonder if the non-executable bounding box was the problem, though it would be incorrect of microsoft to 'tighten up' its parser in this way. You can check by changing the /FontBBox line in one of the .pfb files (it's near the beginning) and seeing if that changes the availability of the font. E.g. the old (6.0) font release of n022003l.pfb has: /FontBBox {-12 -237 650 811} readonly def while the new (8.11) release has: /FontBBox [-60 -273 660 811 ]readonly def That is, curly vs. square braces. Would be helpful to know if just this change makes a difference for Windows XP.
*** Bug 687120 has been marked as a duplicate of this bug. ***
There are significant differences between Adobe fonts and the corresponding Ghostscript fonts causing rendering and PDF generation issues. See bug 687501 for details.
We need to check the metrics against the adobe fonts. See bug 687501.
Valek Filippov released a new version of URW fonts. It is time for new GS font release. No? From the Changelog: 2004-06-06 Valek Filippov <frob@df.ru> Applied 2 glyph changes made by Mashrab. Slightly corrected Vietnamese glyphs autogenerated by fontforge, authogenerated some Vietnamese glyphs for 'Chancery'. Microupdates in non-russian cyrillics here and there. All fonts regenerated form fontforge-2004jun01. 2004-05-31 Valek Filippov <frob@df.ru> Changed EM for Nimbuses Mono back to 775. Added Vietnamese glyphs form urwvn. Thanks to The Thanh Han (the Author), pclouds (who ask Owen about distributing urwvn in the RH distros) and Owen Taylor, who pointed them to me. Some Vietnamese glyphs in NimbusSansCond were autogenerated by FontForge coz it were loosed in the urwvn. Applied Mashrab changes/additions for Nimbuses and Palladio. Mashrab correct me that the glyphs he made/improved are not only Uzbek/Tajik but also shared by Tartar and some other cyrillic-based alphabets. All fonts except Symbol and Dingbats were regenerated from FontForge-2004May23. 2004-04-04 Valek Filippov <frob@df.ru> Fixed FullNames for NimbusRoman No9 (Medium -> Bold), fixed width for NimbusMono (775 for every glyph, coz I scale it to EM == 1000). 2004-04-03 Valek Filippov <frob@df.ru> Fixed bug introduced in NimbusMono with prevoius update. Thanks to Mashrab who point me to it. 2004-02-14 Valek Filippov <frob@df.ru> Added some Uzbek/Tajik cyrillics to Nimbuses from Mashrab Kuvatov <kmashrab@sat.physik.uni-bremen.de> Fixed most of FamilyName/Weight bugs with the great help of VSU (aka Vlasov Serge). Fixed size of cyrillics in the NimbusMono.
Created attachment 726 [details] match.tar.bz2 This is a smple PS program (and a generated PDF file) that shows the difference between Helvetica and the corresponding GS font. Now we just need to fix the glyphs that are too different from the reference implementation.
We discovered from bug 687297 that the NimbusSanL-Regular degree symbol has incorrect metrics. This should be fixed as part of verifying all the metrics match the adobe fonts. AFM line for NimbusSanL-Regu C -1 ; WX 606 ; N degree ; B 151 383 454 686 ; AFM line for Helvetica C -1 ; WX 400 ; N degree ; B 54 411 346 703 ;
*** Bug 687501 has been marked as a duplicate of this bug. ***
Other metrics problem with the current font release: * URWChanceryL is to small. Excerpts from the afm: C 97 ; WX 350 ; N a ; B 72 -11 410 343 ; vs C 97 ; WX 420 ; N a ; B 87 -13 492 411 ; for the original URW fonts. * "germandbls" in NimbusSansL Regular and Oblique is to narrow (easily seen in Alex Cherepanov's match.pdf) or from the afm: n019003l.afm: C 251 ; WX 611 ; N germandbls ; B 126 -20 566 729 ; n019023l.afm: C 251 ; WX 611 ; N germandbls ; B 126 -23 655 729 ; Helvetica.afm: C 251 ; WX 611 ; N germandbls ; B 67 -15 571 728 ; Helvetica-Oblique.afm:C 251 ; WX 611 ; N germandbls ; B 67 -15 658 728 ; AFAICT both these problems are fixed in Filippov's current fonts.
See http://ghostscript.com/pipermail/gs-devel/2004-October/003102.html for another description of the problems with the metrics.
*** Bug 687886 has been marked as a duplicate of this bug. ***
see also bug 688258 for some missing characters a customer would like.
*** Bug 688298 has been marked as a duplicate of this bug. ***
May be it is time to reevaluate the fonts by Filippov? In his new version he claims to reset metrics of latin glyphs to original, and afmdiff shows that he has done that fot the most part ([tT]commaaccent and [uU]tilde are exceptions). IMHO, a new ghostscript font release can be made from ftp://ftp.gnome.ru/fonts/urw/release/urw-fonts-1.0.7pre41.tar.bz2 with URWChanceryL-MediItal taken from ftp://ftp.gnome.ru/fonts/urw/release/urw-fonts-1.0.7pre40.tar.bz2, since in that font cyrillic part has been dropped in the last version. Sincerely, Michail
I'm going to check metrics and fix bugs this one depends on. We discussed it with Ralph and agreed to drop VN (vietnamese) glyphs from my version of urw-fonts for a snapshot release.
I checked metrics against Adobe Core 14 afms with help of afmdiff.awk. Nor gs-fonts-6.0, nor 8.11 nor 1.07pre41 match with Adobe. I count only differences in the original (gs-fonts-6.0) part of fonts. So I would like to ask what do you want to be done with that. E.g. I can check and fix any differences (if any) between 6.0 and 8.11/1.07pre41.
1.0.7pre43 was released. ftp://ftp.gnome.ru/fonts/urw/release/urw-fonts-1.0.7pre43.tar.bz2 Metrics have been checked against gs-fonts-6.0. Non-russian cyrillic glyphs were 'fixed or removed'.
The fonts listed in #24 are not hinted, so until they do, they are not usable yet.
How about solving the TeX Gyre font licensing issue? After all, the TeX Gyre developers are taking the GPL Ghostscript fonts and have already done most of the development works, except they are relicensing the fonts under incompatible terms without showing any authorizations (See bug 689431)!
*** Bug 690099 has been marked as a duplicate of this bug. ***
*** Bug 690269 has been marked as a duplicate of this bug. ***
*** This bug has been marked as a duplicate of 688557 ***
Please disregard the comment #29. Reopening the bug.
Many distros are no longer packaging the gsfonts as released on SourceForge, which have not seen any release in over six years, but rather those found on ghostscript's svn repo: http://svn.ghostscript.com/ghostscript/tags/urw-fonts-1.0.7pre44/ The urw-fonts package on Fedora for instance is based on these. As the packager of "gsfonts" on Arch Linux, I would really appreciate being able to fetch the fonts from an official release instead of using the source tarball provided by Fedora. Any chance this can happen soon? Some relevant links: https://bugs.freedesktop.org/show_bug.cgi?id=18126#c16 http://bbs.archlinux.org/viewtopic.php?id=93521 http://bugs.archlinux.org/task/10593 Thanks in advance for considering this!
Created attachment 6146 [details] bug687297.tar.bz2 This is an improved version of match.ps and generated PDF files that compare Ghostscript fonts with corresponding Adobe fonts. There are glyphs in every font (except Courier) that have different metrics. Some of the glyphs look quite different from the original. The quality of contributed Cyrillic glyphs is visibly worse than the quality of Latin glyphs.
Enhancement still missing in Ghostscript 9.03
The font differences must be generated with the pristine URW fonts (not cyrillic modified) before sending to URW. We should also have a list of glyph that need to be fixed. Many of the glyphs rendered in the PDF files show minor acceptable difference. I don't think URW will want to wade through all the characters.
Partially fixed by reverting to the pristine URW fonts. This removes the problematic Cyrillic glyphs, and gives us a "clean" base to get back to URW with remaining issues.
The latest release from URW+ resolves several issues, I'm closing this bug as it is, frankly, too old and rambling to be useful now. Problems arising with the revised font set should get new bugs.