CP866 and 1.0.0

Matthias Andree matthias.andree at gmx.de
Wed Nov 30 00:44:13 CET 2005


David Relson <relson at osagesoftware.com> writes:

> The explanation is valuable and appreciated.  This sounds like a build
> problem to me.

If you consider configure missing crucial facts about the host system to
be "build problems", then yes, it's a build problem.

The configure[.ac] script currently boldly assumes everyone could do
Unicode, which isn't true. The API is there, it's the same, it works,
but it doesn't support conversion in this particular case -- the input
character set needs to be called "8859-1" rather than "ISO-8859-1" on
older Solaris versions. OTOH, GNU iconv wants 8859_1 not 8859-1...

I see no unison across systems here, and I'm not willing to translate
charset names to extract the iconv codeset name of the day. I'd rather
use --disable Unicode on systems that don't support the IANA names as
from_charset when to_charset is UTF-8.

> In this case, the problem seems to be an inadequate implementation of
> iconv.

Perhaps. Actually it's implementation-defined behavior we're tripping
across.

> We could add the old solaris version to the "not supported" list, but
> that seems excessive.  Why not have configure check for the old
> solaris version and force a non-unicode build?  This would be a pretty
> small change that would contribute to the quality of 1.0, would it
> not?

I'm reluctant to sit on that particular chair.

What we want to know is the answer to: "Does this system support Unicode
in UTF-8 transformation format?". Why should we then ask "Is this system
running Solaris?" What if AIX had the same problem? I don't know, just a
Gedankenexperiment. I'd suggest we try converting a few IANA charsets
such as ISO-8859-1 to UTF-8 and if that fails, we'll assume
--disable-unicode. If someone feels his system inadequately supported
because it has the necessary tables albeit at different names, he can
write the translation code, and if it's good enough, we'll merge it.

The code fragment I posted should be good enough for configure with
minor adjustments. What do you think?

-- 
Matthias Andree



More information about the Bogofilter mailing list