Re: [Bug-XBoard] [bug #33241] xboard quits with "Unable to create font set."

[Tim Mann]

Thanks for the replies.

The first thing to realize here is that because we're using old Xaw widgets and the old X core fonts mechanism, nothing is going to work great. Everyone's effort for years has been to make the new Xft font mechanism work well, convert programs to use it, and make nice fonts that work with it. (BTW, I don't think it's worthwhile thinking about converting the Xaw-based xboard to use Xft. It would require drawing all text to be done in a completely different way and I doubt it's even feasible to do with Xaw widgets. Efforts are better spent on the GTK branch.)

Next... XCreateFontSet is meant to do the kind of thing you've being trying to do by hand or suggesting we write our own code to do. A little background: the "iso8859-n" (for some n) or "iso10646-1" or "koi8-r" etc. at the end of a font name is just an encoding -- it says how character numbers map to glyphs. iso10646-1 means Unicode, while iso8859-n is one of several different 8-bit extensions of ASCII that were invented before Unicode. koi8-r and koi8-u are encodings that were commonly used for Russian and Ukrainian before Unicode. The encoding doesn't tell you for sure what characters are actually in the font, but it's a good guess that (for instance) a font that is encoded in koi8-u will have all the characters you need for basic Ukrainian. iso10646-1 doesn't tell you a lot because the font is always only a subset of Unicode, and you don't know how big a subset it is. In the early days when XCreateFontSet was invented, a lot of X installations didn't have any iso10646-1 fonts, or the ones they had were missing a lot of characters that you could find in some of the other fonts. So XCreateFontSet follows some rules that are encoded in the XLC_LOCALE file for the locale and tries to find a bunch of fonts that will, collectively (and hopefully), have all the characters that you may need in your current locale.

Even though XCreateFontSet doesn't always do a great job, I think it's going to be futile for us to try to hack around it to do something smarter. There are too many languages, too many fonts each with their own set of missing characters and peculiarities, and too many distros that each package the old X core fonts in different ways and let users install different subsets.

With that preface I'll add some point by point comments.

On Mon, May 30, 2011 at 10:39 AM, Arun Persaud <address@hidden> wrote:

Hi

On 05/29/2011 02:04 PM, Tim Mann wrote:
> What does the garbage look like?

screenshot attached

> I think it might be just that you don't
> have a font with the necessary characters.

yes, I think that's the problem...

I think it's an Xaw or Xt bug that it puts up garbage characters instead of the box character. It should "know" that it couldn't find the character it wanted in any of the fonts in the font set.
 

> With the change I pushed, we
> don't show any error messages when XCreateFontSet can't find all the
> font charsets it wants, because for some reason it seems to want a lot
> more than it really needs. (You can see the charsets that are missing
> listed in the xboard.debug file if you run with the -debug flag.)
> Though if you have the
> -adobe-helvetica-medium-r-normal--*-*-*-*-*-*-iso10646-1series, I would
> have expected them to have Ukranian characters.

Unfortunately they just don't. Nobody ever extended Adobe Helvetica to have Cyrillic characters. That's why there is no iso8859-5, koi8-r, koi8-u encoding of helvetica (because they would be missing all the important characters for those encodings) and why the iso10646-1 encoding doesn't have those characters.
 

I tried to load this one, but in the debug file I get message that KOI8
is not supported...

I don't understand that. Maybe that's one of several messages you got and you just thought it was the most important?

As far as I can tell, XCreateFontSet ignores the encoding if you specify it in your pattern, although that doesn't seem to agree with the documentation. For example:

Requested font set for list -*-*-*-*-*-*-14-*-*-*-*-*-iso10646-1
 got list -*-*-*-*-*-*-14-*-*-*-*-*-iso10646-1, locale uk_UA.utf8
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-2
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-3
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-4
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-5
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-koi8-r
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-7
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-9
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-13
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-14
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-15
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-140-jisx0208.1983-0
 got charset -daewoo-gothic-medium-r-normal--14-101-100-100-c-0-ksc5601.1987-0
 got charset -isas-fangsong ti-medium-r-normal--14-101-100-100-c-0-gb2312.1980-0
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-jisx0201.1976-0
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso10646-1


Requested font set for list -*-*-*-*-*-*-14-*-*-*-*-*-total-garbage
 got list -*-*-*-*-*-*-14-*-*-*-*-*-total-garbage, locale uk_UA.utf8
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-2
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-3
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-4
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-5
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-koi8-r
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-7
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-9
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-13
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-14
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-15
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-140-jisx0208.1983-0
 got charset -daewoo-gothic-medium-r-normal--14-101-100-100-c-0-ksc5601.1987-0
 got charset -isas-fangsong ti-medium-r-normal--14-101-100-100-c-0-gb2312.1980-0
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-jisx0201.1976-0
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso10646-1

Requested font set for list -*-*-*-*-*-*-14-*-*-*-*-*-koi8-u
 got list -*-*-*-*-*-*-14-*-*-*-*-*-koi8-u, locale uk_UA.utf8
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-2
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-3
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-4
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-5
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-koi8-r
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-7
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-9
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-13
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-14
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-15
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-140-jisx0208.1983-0
 got charset -daewoo-gothic-medium-r-normal--14-101-100-100-c-0-ksc5601.1987-0
 got charset -isas-fangsong ti-medium-r-normal--14-101-100-100-c-0-gb2312.1980-0
 got charset -misc-fixed-medium-r-normal--14-130-75-75-c-70-jisx0201.1976-0
 got charset -misc-fixed-bold-r-normal--14-130-75-75-c-70-iso10646-1

It gets the same list regardless.
 

I then tried to load some with "-*...-koi8-uni", but
they don't show up in the debug file (I do see them via xlsfonts) and
xboard load the same files as before... perhaps we can trigger on things
like "missing charset KOI8-U" (from xboard.debug) and reset the locale
to english and output a warning in case the correct font can't be found?

Missing koi8-u isn't necessarily a failure. The characters for Ukrainian might be found in an iso10646-1 font (some do have them, though helvetica doesn't), iso8859-5, koi8-r, or others.
 

I also noticed that the xboardrc file shows several lines of font
messages and not just the last one... and the latest ones that I used
are listed on top and therefor when I load the xboardrc file might get
overwritten by the ones below? I haven't looked that much into it yet
though...

That's a feature that HG ported from winboard. The "sizeNNN:" in front says "use this font only for square size NNN".  If you don't have the "sizeNNN:" there, xboard will use the font for whatever the current square size is. But when it saves settings, it saves the fonts only for the current square size. This feature is more important in WinBoard where there is a font selection dialog and you can change the fonts (and the square size too) without exiting and restarting the program.
 

As for other po files: only de tr and uk should work... the other ones
are just are generated by a script from winboard files and will probably
need adjusting by the TP project...

The uk one is missing a lot of translations. The ru, it, and es ones are a lot more complete, though I don't read those languages so I don't know how accurate they are. Maybe we should hack them all up to where they will build and let folks use them. They are probably better than the uk translation already.

BTW: I am thinking we should remove the _( ) from strings that are just debug messages that get printed to debugFP (the xboard.debug file). Two reasons for that: (1) reduce the burden on the translation team -- let them focus on the messages that users really see, (2) xboard.debug is mostly to help us debug problems and it will be easier for us to read files people send us if they are in English.
 

There is a file po/LINGUAS where you need to add the translations you
want make to build... at the moment I only have de tr and uk in there,
since the other ones resulted in errors.

I did find that and edit it... but your other message gives the step I was missing.

cheers
      ARUN