Re: Encoding of Mercurial meta information

Hi Stephan,

Mercurial supports UTF-8. I made some experiments with UTF-8 in file 
names, message logs and user names. Worked out fine - at least with 
hg-1.3.1, didn't try some thing else.

Haven't check the web-interface yet, though, but I think there was some 
config which mentioned UTF-8. The content-type of the email notification 
seems wrong to me, needs to be fixed.

Ah, and of course, just because hg supports UTF-8 this doesn't mean 
anyone should come up with fancy file names or message logs. Especially 
not with fancy files names. Remember, no one got ever fired for using 
ascii-7 :-). The only place where I can see a certain usefulness for 
UTF-8 is the authorname, so it's ok to use it there but you always risk 
that your name might look mangled on someone else terminal (most Linux 
and Solaris terminals are UTF-8 nowadays, but this is certainly not yet 
standard everywhere).

ISO-8859-1 is so ... well ... 20th century, do not use it anywhere in 
our build environment. Full Unicode is not yet practicable at all.

Regards,
    Heiner

Stephan Bergmann wrote:
> What character set should the characters be drawn from used in the meta 
> information of a Mercurial change set (like the -u user name and -m 
> message)?  Plain 7-bit ASCII, ISO 8859-1, full Unicode?
> 
> I noticed that for example Caolán's name is garbled (with a "?") both in 
> the web front end (e.g., 
> <http://hg.services.openoffice.org/cws/cmcfixes65/rev/3ee4cb3a17dd>, 
> XHTML marked as '<?xml version="1.0" encoding="ascii"?>') and the mails 
> sent to [email protected] (which are marked as "content-type: 
> text/plain; charset=us-ascii").
> 
> (By the way, [email protected] is missing from the list of general 
> mailing lists at <http://www.openoffice.org/mail_list.html>.)
> 
> -Stephan
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
> 
>