Re: [viewvc-users] Error due to filename with accented character while building/updating a commit database using cvsdbadmin

On 11/18/2010 06:27 AM, [email protected] wrote:
> I am trying to build a commit database on a Windows 2003 SP2 server, 
> ActiveState Python 2.5, Cvsnt 2.5.05.3744, ViewVC 1.1.5 I get this 
> error when ViewVC processes a file with an accentd character in its 
> name:
> 
> [sources/IK-STAT/doc/src/Funzionalit ikstat.doc [2 new 
> commits]]Traceback (most
> recent call last):
>   File "cvsdbadmin", line 191, in <module>
>     command == 'update', quiet_level)
>   File "cvsdbadmin", line 88, in RecurseUpdate
>     RecurseUpdate(db, repository, path, update, quiet_level)
>   File "cvsdbadmin", line 88, in RecurseUpdate
>     RecurseUpdate(db, repository, path, update, quiet_level)
>   File "cvsdbadmin", line 88, in RecurseUpdate
>     RecurseUpdate(db, repository, path, update, quiet_level)
>   File "cvsdbadmin", line 92, in RecurseUpdate
>     UpdateFile(db, repository, path, update, quiet_level)
>   File "cvsdbadmin", line 64, in UpdateFile
>     print '[%s [%d new commits]]' % (file, len(commit_list)),
> IOError: [Errno 0] Error
> 
> 
> The actual file name is "sources/IK-STAT/doc/src/Funzionalità 
> ikstat.doc", with an accented a letter. ViewVC displays a related error 
> when I try to access the directory in which the file resides:
> 
> Traceback (most recent call last):
>   File "c:\programmi\viewvc\lib\viewvc.py", line 4317, in main
>     request.run_viewvc()
>   File "c:\programmi\viewvc\lib\viewvc.py", line 397, in run_viewvc
>     self.view_func(self)
>   File "c:\programmi\viewvc\lib\viewvc.py", line 1909, in 
> view_directory
>     file_data, options)
>   File "c:\programmi\viewvc\lib\vclib\ccvs\bincvs.py", line 247, in 
> dirlogs
>     alltags = _get_logs(self, path_parts, entries_to_fetch, rev, 
> subdirs)
>   File "c:\programmi\viewvc\lib\vclib\ccvs\bincvs.py", line 1017, in 
> _get_logs
>     ', found %s' % (file and file.path, filename))
> Error: Error parsing rlog output. Expected RCS file 
> E:\Repository\digint\sources\IK-STAT\doc\src\Funzionalit� ikstat.doc,v, 
> found E:\Repository\digint\sources\IK-STAT\doc\src\Funzionalità 
> ikstat.doc,v
> 
> The "expected file" shows a square symbol instead of the accented a, 
> while the "found file" is correct. If I switch the browser enconding 
> (IE8) from UTF-8 to Windows West Europe (I guess it's Windows 1252), 
> the "expected" file is shown correctly while the "found" one is shown 
> with a different character for the accented a. Thereby I believe there 
> are some codepage issues between ViewVC, rlog.exe and filenames under 
> Windows.
> 
> Is this fixed in a later release? Or is it an open issue?

This may be another manifestation of the general problem reported in
http://viewvc.tigris.org/issues/show_bug.cgi?id=292, which is effectively
that MySQL isn't being used in any sort of i18n-friendly fashion.
Complicating this for CVS repositories is that CVS is locale-agnostic
(locale-ignorant?)  There's no way to reliably figure out what character
encoding is in use for the paths in your CVS repositories, log messages,
etc.  (Which makes normalization to, say, UTF-8 for use in the MySQL
database rather difficult.)

-- 
C. Michael Pilato <[email protected]>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand

------------------------------------------------------
http://viewvc.tigris.org/ds/viewMessage.do?dsForumId=4255&dsMessageId=2682768

To unsubscribe from this discussion, e-mail: [[email protected]].

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)

iEYEARECAAYFAkzlPnUACgkQokEGqRcG/W4zsACgmhQMT16xWTnjAMDeZm5HSoID
Fd0AoML5BsgIBI3Y4Ybc0glAU36R3GV0
=OYH3
-----END PGP SIGNATURE-----