Re: downcase'ing wrt locale in smtpmail.el

Re: downcase'ing wrt locale in smtpmail.el

[Simon Josefsson]

yazicivo@ttnet.net.tr writes:

> Hi,
>
> In smtpmail.el, protocol messages are getting downcased with respect
> to current locale. For instance, it downcases AUTH methods (LOGIN,
> PLAIN, etc.) and then compares them with the related downcased symbols
> '(login plain ...). And in Turkish, a downcased I is ı (dotless i),
> not i. Therefore, comparison fails because of logın != login or plaın
> != plain. IMHO, downcased must be applied using en_US locale.
>
> To reproduce the problem, start emacs with LC_CTYPE=tr_TR prefix and
> try to send a mail via smtpmail package.

I just noticed this report.  Is there a function in elisp that does
downcase in the C locale?  For this use, we only care about ASCII, so a
'tr A-Z a-z' would work.  Do smtpmail.el have to implement a
'downcase-ascii' function itself?

/Simon

Re: downcase'ing wrt locale in smtpmail.el

[Reiner Steib]

On Sat, Jan 12 2008, Simon Josefsson wrote:

> yazicivo@ttnet.net.tr writes:
[...]
>> In smtpmail.el, protocol messages are getting downcased with respect
>> to current locale. For instance, it downcases AUTH methods (LOGIN,
>> PLAIN, etc.) and then compares them with the related downcased symbols
>> '(login plain ...). And in Turkish, a downcased I is ı (dotless i),
>> not i. Therefore, comparison fails because of logın != login or plaın
>> != plain. IMHO, downcased must be applied using en_US locale.
[...]
> I just noticed this report.  Is there a function in elisp that does
> downcase in the C locale?  For this use, we only care about ASCII, so a
> 'tr A-Z a-z' would work.  Do smtpmail.el have to implement a
> 'downcase-ascii' function itself?

Chong Yidong has fixed some similar issues in smtpmail.el using
with-case-table / ascii-case-table, see
http://thread.gmane.org/gmane.emacs.gnus.general/64489/focus=64496

See <http://thread.gmane.org/gmane.emacs.devel/68859> for the quite
long discussion about this problem (Subject: ... Locale Dependent
Downcasing in smtpmail).  Oh, know I notice that
<yazicivo@ttnet.net.tr> reported this problem, so I'd guess it has
been fixed by Chong Yidong's change.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: downcase'ing wrt locale in smtpmail.el

[Richard Stallman]

    I just noticed this report.  Is there a function in elisp that does
    downcase in the C locale?  For this use, we only care about ASCII, so a
    'tr A-Z a-z' would work.  Do smtpmail.el have to implement a
    'downcase-ascii' function itself?

You can switch to the standard case table for this.

Re: downcase'ing wrt locale in smtpmail.el

[Simon Josefsson]

Reiner Steib <reinersteib+gmane@imap.cc> writes:

> On Sat, Jan 12 2008, Simon Josefsson wrote:
>
>> yazicivo@ttnet.net.tr writes:
> [...]
>>> In smtpmail.el, protocol messages are getting downcased with respect
>>> to current locale. For instance, it downcases AUTH methods (LOGIN,
>>> PLAIN, etc.) and then compares them with the related downcased symbols
>>> '(login plain ...). And in Turkish, a downcased I is ı (dotless i),
>>> not i. Therefore, comparison fails because of logın != login or plaın
>>> != plain. IMHO, downcased must be applied using en_US locale.
> [...]
>> I just noticed this report.  Is there a function in elisp that does
>> downcase in the C locale?  For this use, we only care about ASCII, so a
>> 'tr A-Z a-z' would work.  Do smtpmail.el have to implement a
>> 'downcase-ascii' function itself?
>
> Chong Yidong has fixed some similar issues in smtpmail.el using
> with-case-table / ascii-case-table, see
> http://thread.gmane.org/gmane.emacs.gnus.general/64489/focus=64496
>
> See <http://thread.gmane.org/gmane.emacs.devel/68859> for the quite
> long discussion about this problem (Subject: ... Locale Dependent
> Downcasing in smtpmail).  Oh, know I notice that
> <yazicivo@ttnet.net.tr> reported this problem, so I'd guess it has
> been fixed by Chong Yidong's change.

I'm happy to hear that it has already been fixed.  I didn't notice any
follow-ups on the post to emacs-pretest, that's why I asked again.

Thanks,
/Simon