Re: Reworking the client and server access method code

[minyard]

address@hidden (Greg A. Woods) writes:

> [ On , July 25, 2001 at 18:29:15 (-0500), address@hidden wrote: ]
> > Subject: Reworking the client and server access method code
> >
> > I've been digging around in the CVS server and client code to add an
> > ssl based server method, and I couldn't help notice what a mess it is.
> 
> Indeed it is.
> 
> > I was curious if anyone was working on cleaning it up.  If not, I'd be
> > willing do clean it up and make the access methods more modular and
> > get rid of most of the ifdef and if statements.  I would also propose
> > pulling each access method into its own file and making buffer-based
> > I/O setup by the access method itself.
> 
> You'd be better off simply ripping everything to do with special custom
> protocols out, including cvspserver, and simply keeping the basic method
> of invoking an external rsh-like tool.  Any protocol you wish to use can
> be done with a simple rsh-like tool, including SSL.

I don't think that's a very good idea.  Even though pserver is not
that great from a security point of view, it's probably sufficient for
a lot of people.  I'm sure people still use Kerberos, and fork is
still a useful debugging tool.

And even if you ripped everything out, you would want a framework to
be able to plug different I/O methods onto the front of "server"; it
would need to sit in basically the same place.

Anyway, my offer still stands to the maintainers of CVS.

-Corey