[dev] Re: [cvs] commit: whups css.php cvs.php diff.php whups/config html.php.dist .cvsignore conf.php.dist whups/graphics deleted.gif whups/templates diff_header.inc diff_request.inc diff_rev.inc dir_back.inc dir_directory.inc dir_file.inc dir_header.inc ...

Anil Madhavapeddy anil@recoil.org
Thu, 21 Dec 2000 13:13:07 +0000


On Thu, Dec 21, 2000 at 12:39:26AM -0500, Chuck Hagenbuch wrote:
> Quoting avsm@horde.org <avsm@horde.org>:
> 
> >   Convert the whole of whups to use CSS.  Some duplication from Horde
> >   here (I gratuitously ripped css.php), but it doesn't have any need to
> >   have Horde installed, and the format used is the same, so we can fall
> >   back easily if needed.
> 
> I can certainly understand this desire. How set on this are you, though? And
> do you see it lasting as whups becomes more complex?
> 

I can't really see it lasting at all.  I was hoping that the 
CVS viewer would just be a simple script that anyone could install
without Horde as well, but more and more feature ideas are coming
to mind that could really use Horde.

a) MIME support:  I'll like to detect the type of a file in 
   the repository a bit more intelligently, rather than just give
   a huge set of extensions => mime type mapping.

b) DB Cache: the viewer is an incredibly good target for caching,
   given that RCS files in a repository change relatively rarely,
   and parsing them is very expensive.  Just checking the file
   timestamp and serializing the CVSLib_file object would speed
   up something like the IMP repository a _lot_.

c) CSS: as above, one of the advantages of Horde is a nice 
   framework for all this.

(Point b is probably taken care off by DB:: in PEAR now)

Now, given that WHUPS will certainly require Horde sooner or
later (as we'd hope, given that it's a Horde module!), we
could either split off the CVS viewer into a separate module
of its own, and release it as a simple script, or just integrate
it fully now.

Horde is a hell of a lot simpler to install these days since
we lost PHPLIB, so I don't think it's _that_ much of a problem,
except for 'hard to install' reputation we've garnered, but
we'll just have to shake that off with IMP 2.4 :)

Anil