[cvsnt] Re: Suggestion for improvement of sspi protocol handling

[Tony Hoyle]

On Thu, 24 Oct 2002 13:59:16 +0000, Bo Berglund wrote:

> Since there are problems with the -x flag for encryption with sspi
> when the flag is added for other protocols I would suggest that cvs
> will just disregard this flag if it does not understand it.
> Right now cvsnt throws an error that stops the execution of the
> command if the -x flag is added (for example through the .cvsrc file)
> and the protocol is *not* sspi.
> This happens in already checked out sandboxes using the pserver
> protocol.

That's a bit dangerous, though, if you've asked for encryption and for
some reason don't get it.  Trusting an unencrypted connection could
conceivably have serious consequences (and you can't force it at the
server side yet...).

One of these days I'll get around to writing a 'cvs capabilities' command
so frontends can find these kinds of things out easily.  I just spent the
weekend fighting with hardware though so maybe not just now... :-)

Tony