http://bugs.winehq.org/show_bug.cgi?id=10952
Summary: Wine prefers Marlett instead of a legible font Product: Wine Version: 0.9.52. Platform: PC URL: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=458234 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: wine-gui AssignedTo: wine-bugs@winehq.org ReportedBy: ovek@arcticnet.no
Currently, Debian discourages installing non-free stuff such as the Microsoft core fonts. Debian policy forbids me from recommending it. (They also consider the Liberation fonts undistributable.) Hence, you have to expect Wine users from Debian to not have fonts like Arial available. The problem is that, if an application asks for such a nonexistent fonts (very common), Wine frequently chooses Marlett, which is fairly illegible. It'd be much more user-friendly if it chose Tahoma or something.
I have a bug report that suggests that whether Wine chooses Marlett or not depends on the fontforge version that Marlett.ttf is generated with (the fontforge in Debian Stable (etch) works, but the one in Debian Testing/Unstable does not), but I don't have much time to investigate why. Perhaps someone at Wine knows. Is it something that should be fixed in the marlett.sfd file, or should the bug be reassigned to fontforge?
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|normal |major
http://bugs.winehq.org/show_bug.cgi?id=10952
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|major |enhancement Component|wine-gui |wine-misc
--- Comment #1 from Vitaliy Margolen vitaliy@kievinfo.com 2007-12-29 18:06:35 --- Don't see a major problem here. Remove all fonts form windows then look at it...
They also consider the Liberation fonts undistributable.
That's too bad for all Debian users. Looks like that distro will be unsuitable for Wine use.
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|enhancement |major
--- Comment #2 from Ove Kaaven ovek@arcticnet.no 2007-12-29 18:54:29 --- That's not helpful. This is a regression that affects the usability of Wine on a free system, and is it not the idea of Wine to be free? I seem to recall something about Wine switching to LGPL for such reasons. (It may or may not be a regression in Wine per se, but it's definitely a regression seriously impairing a fundamental part of Wine, something that is supposed to have worked for years. Which is certainly a major problem. Not a "wishlist item".)
And if you uninstalled Arial from Windows, do you really think Windows would substitute with Marlett instead of, say, Tahoma, or MS Sans Serif, or Bitstream Vera (if you've got that; I do), or whatever else that's at least in a non-symbol character set? What if Microsoft Word opened a document written with the "Dragonmaster" font, and the user didn't have it, would Windows choose Marlett then, or would it prefer a non-symbol font? So, what's better, fixing a deficiency in Wine's font selection algorithm (or maybe in the font itself, if the "this is a symbol font" information is lost, or in fontforge) so that there's a sensible fallback (or maybe even adding font alias configuration to winecfg), or considering an entire, popular distro unsupported, for absolutely no good reason?
(And yeah, I've tried to defend msttcorefonts, but you should have seen the other people's arguments - they say that Wine should remain supported in Debian even without those fonts, because it can still be used to run free Windows applications that aren't written to depend on proprietary fonts, and whatnot...)
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #3 from Jesse Allen the3dfxdude@gmail.com 2007-12-29 19:13:14 ---
They also consider the Liberation fonts undistributable.
That's strange. I guess you'll have to point out what fonts you would expect wine to use, because I have no idea because I don't run debian.
Also I have not had issues with wine's fonts except when I had a broken fontforge setup. I looked at the debian bug, and sure enough, your reporter indicates the problem may be caused by fontforge. My fontforge is from 20060822, so I guess I could try to upgrade it and see if it breaks.
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #4 from Ove Kaaven ovek@arcticnet.no 2007-12-29 20:11:24 --- Exactly which font Wine resigns itself to isn't that important, it's the font's charset that matters most. But since you ask, Wine bundles Tahoma, it could use that, for example. Of system fonts, Bitstream Vera is also a good choice, if available. Or good old Helvetica. The best would probably be to make this user configurable in winecfg; Wine can already substitute font names, it just needs a user interface for it.
But I'm sure an acceptable solution only takes a couple of lines of code somewhere - with the ugliest solution being explicitly excluding Marlett from the font selection algorithm, and settling for anything else, if not explicitly asked for. Better would be to exclude all symbol fonts unless they're explictly asked for, if nothing else helps. Of course, predictability or configurability would be nice, but *that* would be more of a feature request than a major issue, so not something I'm asking for right now.
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #5 from Ove Kaaven ovek@arcticnet.no 2007-12-29 20:55:04 --- As for the Liberation fonts, it may seem strange, but if you're curious, it's basically like this: The license of the Liberation fonts is standard GPLv2 plus an extra restriction or two. The text of the GPL says that you can't add non-GPL restrictions of your own, they'd violate the GPL. The FSF does permit modified licenses based on GPL, but *only* if they change the name of the license to something other than GPL, which Red Hat hasn't done. Thus, the whole Liberation font license is invalid, and Debian has no legal right to redistribute. But Red Hat does not acknowledge the problem, nor does anyone else, hence it doesn't get fixed.
Messy...
http://bugs.winehq.org/show_bug.cgi?id=10952
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|major |minor Status|UNCONFIRMED |RESOLVED Resolution| |INVALID
--- Comment #6 from Vitaliy Margolen vitaliy@kievinfo.com 2007-12-29 20:56:31 --- If it's a regression - run regression testing and find the patch. If your distro is such bent on free - why do you need Wine? It runs apps from not free system! Don't use it and tell all Debian users, that they can't run anything that's not free.
Anyway, you either install fonts, or show the patch that you saying introduced the regression.
Oh and while you at it, make sure that you compiled Wine from GIT and not using broken Debian binaries. And attach output of configure --verbose to verify your system setup is correct.
But I'm sure an acceptable solution only takes a couple of lines of code somewhere - with the ugliest solution being explicitly excluding Marlett from
Patches are welcome.
Closing invalid. User does not have required fonts.
http://bugs.winehq.org/show_bug.cgi?id=10952
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #7 from Vitaliy Margolen vitaliy@kievinfo.com 2007-12-29 20:57:19 --- Unless you show that you have all the required fonts installed (same as on windows) this bug is invalid.
Closing.
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|minor |major Status|CLOSED |UNCONFIRMED Resolution|INVALID |
--- Comment #8 from Ove Kaaven ovek@arcticnet.no 2007-12-29 22:05:10 --- The bug heading is "Wine prefers Marlett". The heading is not "Wine doesn't display Arial". Therefore, the premise of the bug is not invalid. You didn't answer this: "What if Microsoft Word opened a document written with the "Dragonmaster" font, and the user didn't have it, would Windows choose Marlett then, or would it prefer a non-symbol font?" So, why would *you* want to let Wine choose Marlett instead of Arial in this case (which it might, regardless of whether msttcorefonts was installed), knowing that Windows wouldn't do this? Do you *want* to let clear bugs go hidden and unfixed by saying "well, don't do that then", or doing a "the problem doesn't exist, go away" Jedi mind trick? Is that supposed to be how quality assurance works in the FOSS world of "let's get things right under the hood, let's make it solid, let's follow principle of least surprise and make it work on a wide range of systems, because we have no bosses that force us to take shortcuts", that made Linux itself so popular? Alexandre has always had this "get-it-right-at-any-cost" attitude, which we should thus be adhering to here since he's the boss, while your "get-it-working-at-any-cost" attitude would be more appropriate for a commercial product like CodeWeavers CrossOver (and probably work pretty well there - it's just that it's not the same thing).
Yes, definitely, finding where the regression is, and submitting patches, is something that needs to be done in short order, of course, but it's better done with the bug open rather than closed. That's just how bug tracking systems are generally used. Bugs are closed *after* patches are applied, not before. You don't have to be involved in it, either; someone else, who might actually have some interest in this, could do all the work. (Maybe even me, if I find time.) You can just sit back and let it happen. I'll even cloes the bug myself if we find out that fontforge is the problem.
And finally, since you ask, I don't personally have a clue why we need Wine in Debian. I've mentioned Debian's arguments - that you can, in principle, run DFSG-free Win32 software (from SourceForge or whatever) on it. Other than that, well... Do you want me to pull it from Debian? I wouldn't really mind, especially if this is the attitude I get. Should we go to wine-devel and discuss doing so? (Note that even if I do pull it, this bug is still valid. Wine prefers Marlett. You're just hiding it by using msttcorefonts, even on other distros (at least once they upgrade their fontforges). So don't close it before we've come up with a patch or something.)
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #9 from Jesse Allen the3dfxdude@gmail.com 2007-12-29 22:51:31 --- Whatever happens, with no change to wine, and using newer fontforge, you get marlett instead of maybe tahoma. Sounds like you need to research what fontforge is doing.
Also, what makes you think wine is preferring marlett? The only thing that is different is fontforge.
Wine actually prefers Arial from what I understand, and this was a problem until tahoma was created see, bug 8983, and this patch: http://source.winehq.org/git/wine.git/?a=commitdiff_plain;h=7495d814954420c1...
With tahoma and a good fontforge, you should get what you want, afaict.
http://bugs.winehq.org/show_bug.cgi?id=10952
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|major |trivial Status|UNCONFIRMED |RESOLVED Resolution| |INVALID
--- Comment #10 from Vitaliy Margolen vitaliy@kievinfo.com 2007-12-29 22:58:22 ---
answer this: "What if Microsoft Word opened a document written with the "Dragonmaster" font, and the user didn't have it, would Windows choose
It's up to Word what it's doing, not windows.
I still fail to see where the problem is? You claim you do not have any fonts - so you get some garbage displayed. Everything else you said is irrelevant.
Again closing as there are no bugs that you could demonstrate. Not even a specific program that you have this problem with. Read how to properly submit bugs. Just bunch of hypothetical "what if"s not enough for the bug,
http://bugs.winehq.org/show_bug.cgi?id=10952
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dank@kegel.com
--- Comment #11 from Dan Kegel dank@kegel.com 2007-12-30 01:10:46 --- Hey Ove, if you can boil the problem down a bit better, let's discuss it via email, and then maybe open a new bug report. This one has sort of run away.
http://bugs.winehq.org/show_bug.cgi?id=10952
Robert Millan rmh@aybabtu.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |rmh@aybabtu.com
--- Comment #12 from Robert Millan rmh@aybabtu.com 2007-12-30 06:53:22 --- Hi Ove,
I wouldn't waste my time trying to explain Debian's commitment to freedom to Vitaliy. I don't think he wants to understand it.
Why not just focus on the bug instead.
http://bugs.winehq.org/show_bug.cgi?id=10952
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dmitry@codeweavers.com
--- Comment #13 from Dmitry Timoshkov dmitry@codeweavers.com 2007-12-30 07:53:14 --- dlls/gdi32/tests/font.c,test_nonexistent_font() shows that Wine as well as Windows chooses Arial for a non-existent font. If Arial is not available Wine should choose builtin Tahoma. If the latter is not the case you need to investigate why. Perhaps that's due to a broken Fontforge version, perhaps something else.
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|trivial |major Status|RESOLVED |UNCONFIRMED Resolution|INVALID |
--- Comment #14 from Ove Kaaven ovek@arcticnet.no 2007-12-30 10:36:23 --- Vitaliy: I have fonts. Nobody said I didn't have fonts. You know I have at least one font because Wine bundles Tahoma. Of sans-serif fonts, on my minimal-install of Debian, I also have Helvetica, Lucida, Bitstream Vera, ClearlyU, and DejaVu Sans. All great fonts. Of serif fonts, I have Bitstream Charter, Bitstream Vera Serif, Charter, DejaVu Serif, and New Century Schoolbook. Many of these fonts also provide monospace variants, so that's covered too.
Jesse Allen: exactly, but apparently the "good fontforge" is an *old* fontforge. You can't expect people to have to manually downgrade their fontforges forever (and I can't make Debian autobuilders do this at all, they'd use the newest available), you have to find the reason and do something about it.
Dmitry: Yes, that's how it should work, and how it used to work. But something broke it, and I opened this bug to raise awareness of it and find a solution.
Dan Kegel: I don't have the resources to do so yet. I'd love to look into it, but all I can say right now is that fontforge 0.0.20071002 (FontForge 2.0) causes Wine to think Marlett is a better font than Tahoma (or any of my system fonts) for displaying text, while this is not the case with 0.0.20061019 (FontForge 1.0). I could provide my VMware virtual machine, if that'd help... or I could make traces, if you know what to look for. For example, here's a diff of traces from running winecfg (after aliasing MS Shell Dlg to Arial instead of Tahoma), using WINEDEBUG=+font:
@@ -74,7 +74,7 @@ trace:font:ReadFontDir Found "marlett.ttf" in "/usr/lib/../share/wine/fonts/" trace:font:AddFontToList Loading font file "/usr/lib/../share/wine/fonts//marlett.ttf" index 0 trace:font:get_familyname Got localised name L"Marlett" -trace:font:AddFontToList fsCsb = 80000000 00000000/00000000 00000000 00000000 00000000 +trace:font:AddFontToList fsCsb = 80000001 00000000/00000001 10000000 00000000 00000000 trace:font:AddFontToList Added font L"Marlett" L"Regular" trace:font:ReadFontDir Found "tahomabd.ttf" in "/usr/lib/../share/wine/fonts/" trace:font:AddFontToList Loading font file "/usr/lib/../share/wine/fonts//tahomabd.ttf" index 0 @@ -2047,7 +2047,7 @@ trace:font:DumpFontList Family: L"System" trace:font:DumpFontList L"Regular" 00000001 16 trace:font:DumpFontList Family: L"Marlett" -trace:font:DumpFontList L"Regular" 80000000 +trace:font:DumpFontList L"Regular" 80000001 trace:font:DumpFontList Family: L"Tahoma" trace:font:DumpFontList L"Bold" 20000013 trace:font:DumpFontList L"Regular" 20000013 @@ -2167,50 +2167,42 @@ trace:font:WineEngCreateFontInstance L"MS Shell Dlg", h=-11, it=0, weight=400, PandF=00, charset=0 orient 0 escapement 0 trace:font:WineEngCreateFontInstance not in cache trace:font:WineEngCreateFontInstance substituting L"MS Shell Dlg" -> L"Arial" -trace:font:WineEngCreateFontInstance (it=0, bd=1) is selected for (it=0, bd=0) trace:font:WineEngCreateFontInstance (it=0, bd=0) is selected for (it=0, bd=0) -trace:font:WineEngCreateFontInstance Chosen: L"Tahoma" L"Regular" (/usr/lib/../share/wine/fonts//tahoma.ttf/(nil):0) -trace:font:OpenFontFace "/usr/lib/../share/wine/fonts//tahoma.ttf"/(nil), 0, 0 x -11 +trace:font:WineEngCreateFontInstance Chosen: L"Marlett" L"Regular" (/usr/lib/../share/wine/fonts//marlett.ttf/(nil):0) +trace:font:OpenFontFace "/usr/lib/../share/wine/fonts//marlett.ttf"/(nil), 0, 0 x -11 trace:font:WineEngGetFontData font=0x122078, table=VDMX, offset=0x0, buf=0x33f668, cbData=0x6 ...
So, does this point to a fontforge problem so I can send the report there and get this over with, or do I have to keep circumventing obnoxious bug closers here for a while? I can make a duplicate bug if desired.
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #15 from Dmitry Timoshkov dmitry@codeweavers.com 2007-12-30 11:14:47 --- (In reply to comment #14)
trace:font:DumpFontList Family: L"Marlett" -trace:font:DumpFontList L"Regular" 80000000 +trace:font:DumpFontList L"Regular" 80000001
This one looks like the problem. For some reason Marlett has bit 1 set in fsCsb[0] of the fontsignature. Sounds like a Fontforge bug, since fonts/marlett.sfd has 'Encoding: Symbol'.
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |FIXED
--- Comment #16 from Ove Kaaven ovek@arcticnet.no 2007-12-30 11:46:54 --- Allright, thank you. I guess I'll reassign the report to fontforge and let them figure it out.
http://bugs.winehq.org/show_bug.cgi?id=10952
James Hawkins truiken@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|FIXED |INVALID
--- Comment #17 from James Hawkins truiken@gmail.com 2007-12-30 12:20:05 --- It's invalid, not fixed.
http://bugs.winehq.org/show_bug.cgi?id=10952
James Hawkins truiken@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #18 from James Hawkins truiken@gmail.com 2007-12-30 12:20:56 --- Closing.
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #19 from Dmitry Timoshkov dmitry@codeweavers.com 2008-01-09 00:42:05 --- Ove, looks like there is a duplicate bug report in the Debian buzilla:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=451827
Since your report contains more details about the problem an older one probably can be closed.
http://bugs.winehq.org/show_bug.cgi?id=10952
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |Ronny.Standtke@gmx.net
--- Comment #20 from Dmitry Timoshkov dmitry@codeweavers.com 2008-01-09 00:45:27 --- *** Bug 10430 has been marked as a duplicate of this bug. ***
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #21 from Ove Kaaven ovek@arcticnet.no 2008-01-09 06:24:38 --- I saw that other report, yes. It looks quite different from this one. In this one, it's a case of the code ranges being misdeclared, causing Wine to choose the wrong font, but (presumably) display the correct glyphs from it. In that other bug, it looks like a case of choosing the right font, but displaying the wrong glyphs from it (or in that case, one and the same glyph all over the place). It's likely both are related to fontforge issues, but it never quite looked like the _same_ issue to me.
(By the way, Debian's bug tracker isn't a Bugzilla. It's far easier to deal with than a Bugzilla...)
http://bugs.winehq.org/show_bug.cgi?id=10952
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |mizvekov@gmail.com
--- Comment #22 from Dmitry Timoshkov dmitry@codeweavers.com 2008-01-30 23:22:47 --- *** Bug 10660 has been marked as a duplicate of this bug. ***
http://bugs.winehq.org/show_bug.cgi?id=10952
Ove Kaaven ovek@arcticnet.no changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|CLOSED |UNCONFIRMED Resolution|INVALID |
--- Comment #23 from Ove Kaaven ovek@arcticnet.no 2008-02-08 20:28:43 --- The Debian fontforge maintainer has reassigned the bug(s) back to Wine, as apparently the fontforge folks are claiming that this is a Wine problem. (Supposedly the old fontforge behaviour wrt encodings were incorrect, which they've fixed, and hence Wine should also be fixed accordingly, and generate the font using some "sym.ttf" method.) So I'm reopening this bug in this bug tracker here, too... maybe you can mark it as a duplicate of 10660 or something, though, since that one seems like it's still open.
and if you really want to continue your blame-game/screaming-match habits, you can go to the fontforge developers and tell *them* what their problem is...
http://bugs.winehq.org/show_bug.cgi?id=10952
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |DUPLICATE
--- Comment #24 from Dmitry Timoshkov dmitry@codeweavers.com 2008-02-08 20:41:13 --- Then this is a duplicate.
*** This bug has been marked as a duplicate of bug 10660 ***
http://bugs.winehq.org/show_bug.cgi?id=10952
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #25 from Dmitry Timoshkov dmitry@codeweavers.com 2008-02-08 20:41:28 --- Closing duplicate.
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #26 from Dmitry Timoshkov dmitry@codeweavers.com 2008-02-11 02:58:20 --- (In reply to comment #23)
The Debian fontforge maintainer has reassigned the bug(s) back to Wine, as apparently the fontforge folks are claiming that this is a Wine problem. (Supposedly the old fontforge behaviour wrt encodings were incorrect, which they've fixed, and hence Wine should also be fixed accordingly, and generate the font using some "sym.ttf" method.)
Ove, an original marlett.sfd has been created by Transgaming, can you please tell how it was created, particularly how did you specify the encoding? George Williams claims that someone hand-edited the sfd file and added the line "Encoding: Symbol" manually. Is that true?
http://bugs.winehq.org/show_bug.cgi?id=10952
--- Comment #27 from Ove Kaaven ovek@arcticnet.no 2008-02-11 13:43:25 --- I don't know the details. The font was made by David Hammerton, who is no longer employed by TransGaming. You could try to track him down if you really want to know. Though I wouldn't consider it implausible that he'd hand-edit the file.
-
wine-bugs＠winehq.org