Le Fri, May 17, 2002, à 03:59:23PM +0200, Mattam a écrit:
> > * how does it behave when the XSLT processor isn't here? (we need to
> >have dia not depend on the xslt processor, just suggest it). Ideally, we
> >gray out the option; at the minimum, we should display an error dialog box.
> Use dlopen. But I don't know how to avoid compiling it (using #ifdef's ?)
Oh, if you use dlopen, then it's fine: the ELF loader will not complain
loading your plug-in even if the xlst lib is not here. Then, you only have
to fail gracefully if dlopen returns an error.
> > * is the string handling UTF-8-clean ? This is a must. At the
> >moment, there still must be #hell to handle both a non-UTF-8 gtk (*nix) and
> >a modern GTK (Win32).
> Not concerned.
I take this as meaning: there is no string handling at all (except
perhaps concatenations) in your plug-in. OK, this is UTF-8 clean.
> >ObPatchNote: it would be even better if the archive was only a patch (with
> >the new files patched against "empty" (there is an option for that in "cvs
> >diff")
>
> I tried the option (-N) but I got errors from cvs not founding the
> CVS/Entries file inside my local directory (which is quite logical :)
> Could you tell me the exact way how to do it?
Oh, I see.
You're adding a whole new directory (of course). Hmmm. Perhaps a way would
be to have a duplicate tree from the same CVS source, but without your
modifications. Then, I guess a real "diff(1)" (/not/ cvs diff) with a couple
-x arguments (to avoid diff'ing the CVS directories, for instance, or the
.cvsignore'd files) will do the trick.
-- Cyrille
--
Grumpf.