Re: CVS, Linus, and us

Ulrich Windl (Ulrich.Windl@rz.uni-regensburg.de)
Thu, 22 Feb 1996 16:17:44 +0100


On 21 Feb 96 at 8:39, Warner Losh wrote:

[...]
>
> I think that if the goal is to show Linus that CVS is a good thing, it

Well, I think CVS has some flaws, too: If you recognize that you are
importing into the wrong module by mistake, you can hit ^C, but CVS
won't clean up the mess. Doing that is rather ugly. I think CVS would
need a transaction concept here; hitting ^C would undo the last
operation (RCS tries to do that). For large projects, RCS and CVS
talk too much by default (my impression).

> might be better to point to things like, say, Linux/SPARC and other
> parts of Linux that use it. Have those folks rub elbows with Linus at
> conferences and what not. Over time Linus can judge for himself if
> these folks are saving time or wasting it. I don't think a single
> DEMO system would convince me if I were in this spot.

Right!

>
> Also, read only CVS access is easy to setup. However, it is called
> sup :-). Read only anonymous CVS is hard to setup and takes lots of
> babysitting. As does a SUP server. They also take lots of bandwidth.
> The FreeBSD folks were on a T3 and that wasn't enough (OK, it was
> multiplex with ftp.cdrom.com). If something like this is made
> available, I think that you'd need to find a couple of T3 connected
> sites...
>
> Warner

CVS IS the right concept, but the implementation still has some
flaws (And Linux is the right OS, of course...).

Ulrich