[cvsnt-dev] Unicode file names

Torsten Martinsen torsten at vip.cybercity.dk
Sat Jul 29 22:38:02 BST 2006

Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.

Tony Hoyle wrote:

>Torsten Martinsen wrote:
>>I remember reading at some point that CVSNT supports arbitrary file 
>>names, including Japanese etc. But how is this implemented? Can I pass 
>>the filename in UTF-8 encoding?
>If you start the process using unicode APIs (CreateProcessW etc.) then cvsnt 
>gets the filenames as unicode.  
So I can start CVS and specify a filename using a wchar_t string. But 
how about the -F option, used by the cvsgui protocol? From looking at 
the code, it looks as if that expects ANSI filenames only.

>If you also pass --utf8 then it switches the 
>client into utf8 so it's able to store arbitrary unicode characters (I guess 
>you could just pass the command line as UTF8 but I'm not sure how the shell 
>would interpret it since it's expecting ANSI).
>If you do that all the sandboxes with non-ASCII characters in their filenames 
>will need an update and probably a re-checkout, since the entries files get 
>written and parsed as UTF8.

This is only if I specify --utf8, not if I just start CVS using 
CreateProcessW(), right?


More information about the cvsnt-dev mailing list