[gambit-list] Some UTF-8 support issues.

Christian Jaeger chrjae at gmail.com
Mon Sep 7 19:21:14 EDT 2009


> I think that current locale settings (locale(7)) should be used as
> default value.

Software on Unix like systems, at least Linux, tends to use utf-8 for
encoding of filenames nowadays, but there's no global enforcement for
this. The kernel and libc treat paths just as byte strings. I guess OS
X has some enforcement law to make everything use utf-8. Windows afaik
stores the used encoding in the filesystem, which should make it more
clear which encoding to assume (and might even do the transcoding for
you, not sure).

I think you can't really assume locale settings for changing this
behaviour; if you do, you're fine as long as you don't change the
locale for the lifetime of a particular filesystem. Since you're
inevitable also going to deal with filesystems written with different
encoding assumptions sometimes.

Probably there's no other way to deal with it than the same way you
deal with textual file ports: specify the encoding when opening the
port. Meaning, give the encoding when calling |open-directory|. Maybe
have another Gambit startup flag to set the default. (Gambit doesn't
check the locale for file ports, why should it use the locale for
directory ports?)

(One could argue that if the encoding is unknown, input/output
procedures should return u8vector's instead of strings, and accept
u8vectors instead of strings.)

Christian.

(This is my first mail to a mailing list sent from Gmail, going to see
whether lines are too long. If they are, what are other Gmail users
doing about it?)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.iro.umontreal.ca/pipermail/gambit-list/attachments/20090907/05de8e9d/attachment.htm>


More information about the Gambit-list mailing list