Re: Encoding of .subversion config files?

Jim Blandy writes:
> On 3/8/06, Vincent Lefevre <vincent+svn@vinc17.org> wrote:
> > On 2006-03-08 13:38:16 -0800, Jim Blandy wrote:
> > > I believe common practice is to just read shared config files with the
> > > process's current locale's encoding:
> >
> > > if people want to share, they need to use ASCII, or agree on a
> > > common encoding.
> >
> > or the file should contain the encoding (explicitly or implicitly).
> > This is often the case in new applications.
>
> That's fine, but it entails a lot of work.

Yes, but we should make it consitent anyway.

> - What syntax should we use? The files aren't XML, so we need to make
> up our own.

Something like:
- The first line including
  "(en)?coding: ?([a-zA-Z][a-zA-Z0-9-]+)". (Just a suggestion, could
  be discussed further).

> - How shall we name encodings? I recommend we should follow XML's
> naming conventions, to avoid having to think much.

Something like that, yes. Or the same names as --encoding accepts,
which should commonly be the same. (This is what the iconv lib on the
system supports.)

> - Given an encoding name, how shall we perform the conversion from
> that encoding to UTF-8?

Using the APR xlate APIs (uses libiconv or apr-iconv).

> - Who will write the patch?
>
<evil grin>

I don't think the implementation is had per se, but we need to be
careful about compatibility. But since we're already so
inconsistent...

Best,
//Peter

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Mar 9 21:25:23 2006