Re: spurious characters appearing in messages

Pablo Saratxaga (srtxg@linux.chanae.stben.be)
Tue, 1 Jul 1997 03:13:32 +0200 (MET DST)

Kaixo !

> I've been noticing odd characters appearing in messages gated via ifmail.
>
> This is a clip from c_echo, the characters in the left column:
>
> In C, memory is typically divided into 3 spaces, stack, heap, and data. All
> ́automatic variables - i.e. those declared within functions and *not* declared

The "́"... you use the "tx" isn't it ?

The problem is the so-called "soft CR".
There is no such concept as "soft CR" outside fidonet, so when gating
two choices are available: 1) keep the char as is 2) convert to a simple
space.
The solution 2 will cause problems for messages written in russian, japanese,
chinese, any eastern language and several western ones as well, as that code
(0x8d BTW) is used by charset codings for those languages.
In other words the "soft CR" idea isn't a very good one, it only works on
pure ASCII (eg english-only) areas.
So I choose better not to change it, that does less harm (on my node I route
echos/groups using 7 different charsets).
Then when the charset conversion is done that character is converted too.
In the DOS encoding cp437 it corresponds to "́" (igrave) so it is converted
to it when converting to iso-8859-1.

If you want to override that you can either recompile whithout the
-DIGNORE_SOFTCR define in the CONFIG file (note however that for PGP signed
messages the "soft CR" will be kept as for PGP messages no conversion is done
at all, in order to not invalidate the signature), then it will be converted
to espace.

Or you can edit the maptab file for cp437 to iso-8859-1 conversion, search the
line:

141 236

(141 is 0x8d in decimal)
and replace it as you like, eg

141 " "

or simply erasing it (141 will then not change).

PS: All that only applies to the "tx" versions, as the official ones
simply allways convert it to " " (espace).

-- 

A bientôt, Pablo Saratxaga

~ ~ :wq ;-) PGP Key available, key ID: 0x8F0E4975