Re: linux & nihongo

[turnbull@example.com (Stephen J. Turnbull)]

   I am trying to install Japanese capabilities on my linux system.
   I read Craig's html document 'linux & nihongo' with interest. A 
   few remarks & questions:

I haven't read Craig's doc yet, won't have time before I head for the
U.S.  I have a very specific set of needs, so what I say may not be of 
much use to people with different needs, but here goes.

   -Craig said that you have to have kon running before starting X,
    and that otherwise kterm will not work. On my system that is
    not the case. Kterm works also if X is started in the usual way.

I have no problem with kterm with or without kon.  I found kon to be a 
real pain in the neck, actually.  I don't have kinput running,
however.  I do all of my nihongo in mule, and read nihongo pages with
Mosaic L10N.

   -I found that 'joe' can also display Japanese if it is started 
    with the option -asis. With wnn/uum you can also enter Japanese 
    in joe (and also into the 'pico' editor of pine 3.91).

This should be possible with 'most any 8-bit clean program.  For
example, 'flex -8' works fine.  (I wrote a SJIS/JIS/EUC recognizer in
flex; it wasn't very useful and I don't know where it is now.  But it
worked.)  I'm using Canna now, not Wnn, which I like better than the
ancient version of Wnn I use on the Suns at school.  Except for the
stupid[tm] use of ^O to enter henkan mode.  (It's only stupid because
I use ^O a lot for open-line ;-)  But I only enter nihongo in Mule, so 
I don't know how well it interfaces to input programs like kon or
kinput.

   -I have a problem with kon. When I am logged in as root, it starts
    up normally. If I then quit kon and log in as user on another VC, 
    it does not show text (it only shows blank highlighting bars,
    if any highlighting is present on the page).
    I can only run kon as user if I am running it at the same time on
    another VC as root. Does anyone have any idea what causes this?

There seems to be some problem accessing the fonts.  I never could
figure out what it was.  I tried chmod'ing everything in sight and
that didn't help.  YMMV.  What seems to happen in the root + another
VC case is that the new kon recognizes that the code is already
loaded, uses that, which contains references to the fonts that the
root instance loaded into kon's space.  I wonder if there may be
something in JE itself that needs to be chmod'ed; I didn't try that.

I haven't really figured out what the benefit of JE is, in any case,
as long as you're running X.  (Native speakers presumably would like
to put kanji in the names of their files; I have found that this is
problematic.  When I'm traveling, for example, I often take files on a
DOS-format floppy.  If there are kanji in the names, Windowze and DOS
often get very confused about the existence of the files.  They get
listed (at least by 4DOS), but you can't do anything with them.  So I
name directories and files in English or Romaji.  I've noticed that
most everyone on the Web does too.)  What do y'all use JE for?  I
mean, do jperl etc get upset without it?  do regexp searches in Mule
fail on kanji?  Or is it really only for the file system and konsole,
but data manipulations on files and networks are happy as long as it's 
8-bit clean?

This may be because I live in Mule, and only come out to buy
groceries.

I do wish Netscape would do Japanese, although Mosaic L10N works fine.

Steve