Re: [tlug] Problem sort-of solved ( GTK2: Displaying Japanese font names in Romaji)

["Stephen J. Turnbull" <stephen@example.com>]

>>>>> "Matt" == Matt Gushee <matt@example.com> writes:

    Matt> But I don't like it.

I didn't ask you to *like* it. ;-)

    Matt> It's fragile, it doesn't fix what gets auto-generated--
    Matt> *and*, here's the creepy thing, it appears that in certain
    Matt> cases the per-user font cache in my home directory, which
    Matt> gets auto-generated at intervals, can mask the hacked one in
    Matt> the font directory [see also Meta-rant below].

Yes.  All of this is more or less documented behavior.  I'm not sure
why there is a per-user font cache, to be honest, but obviously if it
is used it should mask the system caches.

Unfortunately, what isn't documented is how and why the mojibake names
get into the font configuration caches.  This would appear to have more
to do with FreeType (fc-cache is documented to be a FreeType-specific
tool) and the font foundries than with fontconfig.

    Matt> By the way, it appears that Fontconfig's config files
    Matt> (e.g. /etc/fonts/{fonts|local}.conf) don't affect the
    Matt> behavior of fc-cache.

fc-cache only needs fonts.conf to find out where the fonts are in the
file system.  Then it extracts the font attributes from font metadata
contained in the font file, and serializes them as a fontconfig
pattern.  This is associated with the file name.  Ie, it does not
munge the font attributes in any way.  It's just an directory like the
X11 fonts.dir, saving on file system accesses, which are very
expensive compared to disk space or CPU time.

    Matt> I get frustrated by the growing amount of seemingly random
    Matt> behavior on Linux systems, which emulates one of the most
    Matt> annoying aspects of Windows.

Hey, you should try maintaining a system based on DarwinPorts. :-/

Does http://turnbull.sk.tsukuba.ac.jp/Tools/Attitude/why-i-must-leave-tlug.html
ring any bells here (see section "Stress")?

    Matt> I have to wonder: is this an *inevitable* side effect of the
    Matt> effort to make Linux more "user-friendly?" Or is it due to
    Matt> the particular (Windows-like) model of user-friendliness
    Matt> that is favored by the mainstream? Or what?

I think it's pretty much inevitable, and it has much more to do with
the fact that the user-level stuff is mostly done for the fun of it
than with any of the design or development models.  Face it, user
demands are random, obnoxious, mendo-kusai.  What volunteer in his
right mind is going to spend 10X the effort to date just to get from
90% user satisfaction to 95%, and in process pissing off a vocal 5%
who *were* satisfied until he made the last tweak?  But 90% for free
is good enough to kick the legs out from under any commercial efforts
charging $200/copy.  (Of course, if the reason for getting from 90% to
95% is to kneecap Microsoft, you can afford to give the results away.
Just ask Sun Microsystems. :-)

On the other hand, the system-level utilities are getting to be
*good*.  I set up my first RAID array today, but I spent more time
futzing with the linux/Documentation files than I did Googling for
"Linux RAID HOWTO", reading, and setting up the array (not to mention
emerging mdadm and lvm2 in the process).  Do you think this stark
contrast is an accident?  I don't.

-- 
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.