On Sat, Sep 04, 1999 at 01:12:50PM -0400, Fairlight wrote:
>On Sat, Sep 04, 1999 at 05:57:00PM +0200, J Horacio MG blurted:
>> Fairlight dijo:
>> > On Sat, Sep 04, 1999 at 10:15:21AM +0200, J Horacio MG blurted:
>> > > http://www.mutt.org is «the mother of all sites» ... well, as far as
My version of mutt interprets these characters correctly (by converting them
from iso-8859-1 to iso-8859-13 which I use).
[...]
>> > So why do the 1/2 symbol and the top-right-double-frame symbol show up in
>> > that email...
[...]
>The 1/2 is quoted up there above...the character preceding "the mother of".
>And after "all sites" is just one character...it's the top-right corner of
>the DOS charset's double-outline box.
>
>> I wasn't aware they would bother at all (I thought they could be read by
>> anyone using iso8859-1). As for the "N" with the tilde on top, well...
>> being as it is at the bottom of the message, I presumed it wouldn't be
>> a hassle for anyone.
>
>Nah, not a hassle...a curiosity...I can't figure out why I'm using the same
>charset in mutt, but it displays oddly.
There is no top-right-double-frame in iso-8859-1, so I guess your font
is probably hardware default cp437. And you're probably telling mutt
with a "set charset" that it's iso-8859-1.
>Unless mutt doesn't ACTUALLY
>invoke the charset on the linux console, and I would have to do that with
>setfont or something.
Yes, that's the job of setfont, not mutt. Although console-chars from
the console-tools (http://www.multimania.com/ydirson/en/lct/) package
make setfont obsolete (more or less).
>Anyone clueful on this?
Well, I've been hacking some linux console fonts lately.
>> My apologies, and please let me know if it looks ok now.
>
>[~N] in .sig now...and you needn't have redone it. No apologies
>needed...I'm probably just not understanding how mutt/linux/charsets all
>interact, and would like someone to briefly explain to me where I'm going
>wrong. :)
There are two things that might be wrong:
1) Application Character Map shows which characters written out to the
console map to which Unicode characters. In other words, it is the
active charset.
2) Screen Font Map shows which Unicode characters can be shown with
currently loaded font and in which font positions.
There are three hardcoded ACMs:
1) iso-8859-1 (the default, can be enabled by "ESC ( B" control sequence)
2) vt100 graphics (enabled by "ESC ( 0" control sequence)
3) cp437 (enabled by "ESC ( U"; was "straight to font" in Linux 2.0)
The fourth (selected with "ESC ( K") is user defined and can be changed
with consolechars -m <acm>.
iso-8859-1 is the default (with 2.2.x kernels), so this should be OK.
But there also is the Screen Font Map, which is *not* loaded by default.
Therefore, when a program (e.g. mutt) displays some non-ASCII character,
the kernel does not know it is in a different position in the font
(cp437) and so it displays the wrong one. Solution: load the screen
font map with consolechars -u <sfm>. If you don't have console-tools,
you can use loadunimap from kbd package. However there is no way to
select a custom ACM in this package, so you have to stick with one of
the defaults or use an inconvenient 8-bit translation table that combines
both ACM and SFM in one translation table.
You can read more about this in the documenation that comes with
console-tools.
Marius Gedminas
--
$ fortune
$3,000,000
