Bug 69322 - python ebuild does not support wide char type functions
Bug#: 69322 Product:  Gentoo Linux Version: unspecified Platform: All
OS/Version: All Status: RESOLVED Severity: enhancement Priority: P1
Resolution: WONTFIX Assigned To: python@gentoo.org Reported By: caglar@pardus.org.tr
Component: Core system
URL:  http://www.python.org
Summary: python ebuild does not support wide char type functions
Keywords:  Inclusion
Status Whiteboard: 
Opened: 2004-10-28 14:28 0000
Description:   Opened: 2004-10-28 14:28 0000
Python [ 2.3.x series ] can handle wide chars, but current ebuilds not use this
functions properly. For example in tr_TR.UTF-8 locale, upper case of "i" is
differ than "I" ( correct one is I WITH THE DOT ABOVE ), and lower case of "I"
is differ than "i" ( correct one is LATIN SMALL LETTER DOTLESS I ) but if wide
char support is not enabled then python converts them into "I" and "i"
respectively, which is wrong.

Maybe a local USE flag can be implemented and used in ebuild. ( Patch is
written by Baris Metin <baris@uludag.org.tr> )

------- Comment #1 From S.Caglar Onur 2004-10-28 14:29:54 0000 -------
Created an attachment (id=42802) [details]
python-2.3.4-wctype.patch

written by Baris Metin (baris@uludag.org.tr)

------- Comment #2 From Alastair Tse (RETIRED) 2004-10-29 02:09:49 0000 -------
i'd support adding it as a default option and not underneath a useflag. we'll
consider this for python-2.4

------- Comment #3 From Heinrich Wendel (RETIRED) 2005-04-24 16:14:06 0000 -------
what's the status of this?

------- Comment #4 From Alastair Tse (RETIRED) 2005-04-26 05:13:15 0000 -------
well, upon further investigation, there were soem comments about wctype causing
problems for certain locales, and a proposal to remove that switch for later
versions of python.

so i'm not so sure about adding this switch in now.

https://sourceforge.net/tracker/index.php?func=detail&aid=1076790&group_id=5470&atid=105470
http://mail.python.org/pipermail/python-dev/2004-December/050193.html

------- Comment #5 From S.Caglar Onur 2005-04-26 05:48:05 0000 -------
Because of this problem, i suggested to implement a use flag, so decision may
made by user( all Turkish spoken users need this to correct case operations ).
I think "unicode" flag suitable for this situation like this;

if use unicode; then
        myconf="${myconf} --with-wctype-functions"
fi

what u think?

------- Comment #6 From Alastair Tse (RETIRED) 2005-04-26 06:05:22 0000 -------
unicode isn't a suitable flag in this case, if it is only for the turkish
locale. the problems that they describe seem to be quite serious (toupper() and
tolower() giving the wrong output), but i'm not sure whether it applies to the
turkish locale. 

i'm not quite sure what the status of linguas support is, but maybe using that
to enable it would be better. the unicode is used for many other languages like
chinese and japanese that don't require wctype support for normal operations.

------- Comment #7 From S.Caglar Onur 2005-04-26 06:34:44 0000 -------
After talked with kloeri in irc, i just send an email to python-dev to get
status of wctype support in Python. As soon as they reply, ill also update
here...

------- Comment #8 From Alastair Tse (RETIRED) 2006-09-02 05:03:17 0000 -------
Reference: http://mail.python.org/pipermail/python-dev/2005-April/052968.html

There was never a suitable conclusion to this. One possible solution would
probably be only enabling this for the turkish locale, but I really don't
want to do that.

REOPEN if a suitable solution can be found.