[OpenAFS] Improving collaboration

[Jeffrey Hutzelman]

On Fri, 11 Jan 2002, Ted Anderson wrote:

> On Fri, 11 Jan 2002 12:01:33 -0500 (EST) Jeffrey Hutzelman <jhutz@cmu.edu> wrote:
> > On Fri, 11 Jan 2002, Ted Anderson wrote:
> > > On Thu, 10 Jan 2002 15:52:48 -0500 (EST) Jeffrey Hutzelman <jhutz@cmu.edu> wrote:
> > > > http://www.fsck.com/rtfm/
> > > The RT/FM system seems pretty focused on tracking problems.
> > 
> > Are you sure you're not confusing RT/FM with RT?  The page I pointed
> > you at is an instance of RT/FM; what it happens to contain is the
> > manual for RT.  The two systems have a very similar look, but RT/FM is
> > _not_ RT.
> 
> Yes, you are right.  But I couldn't find any docs on RT/FM itself.

There aren't any, other than what's included in the package itself.  RT/FM
isn't really quite ready for widespread distribution, but I have reason to
believe it might be stable enough for our use.


> > > I looked at a few of the various Wiki engines[1] and there are a lot of
> > > them.  It looks like UseModWiki[2] might be easiest if the web server is
> > > already using ModPerl.  The MoinMoin engine used by Cees' PhotoWiki
> > > seems reasonable too, unless installing Python whould be an extra
> > > hassle.  We could discuss various features if there is interest in that.
> 
> I forgot the link to MoinMoin[1].

That's OK; there's a fairly complete list of implementations at
http://www.c2.com/cgi/wiki?WikiEngines.

We certainly have python available as well, since Mailman is written in
python.  But I'm a bit unclear on whether MoinMoin expects to run as a CGI
or as its own web server.

> > Modperl is certainly available -- RT uses it -- but my investigation
> > suggests that, if a wiki is what people want, TWiki might be a better
> > choice.  For one thing, I feel uncomfortable with locking ourselves into
> > something that doesn't support authentication and a moderately fine
> > grained access control model.
> 
> Well, call me "radically egalitarian"[2]; I think keeping RCS revisions
> of all the pages is good enough.  Having to register to edit pages is a
> bit off putting, but I don't think it is a huge deal.  Certainly, the
> features of the system looks very nice.  The TWikiWebs[3] are arguably a
> better approach to partitioning than the UseMod SubPages[4].  But a
> feature by feature comparison will be difficult without a lot more
> experience than I have.  Anyway, I'm okay with TWiki.
> 
> We may not get a lot of feedback from OpenAFS people until we actually
> try it.  Maybe we should just go for it and see how people like using a
> Wiki.  There is always the risk that if we don't get critical mass the
> system will just atrophy, but that's an unavoidable.

I have no objection to this approach; in fact, I was thinking of just
putting something up and seeing what happens.  The only problem with
experimenting that way is that there will be extra work to do if we end up
deciding to go a different route.  Even so, IMHO something is considerably
better than nothing, which is pretty much what we've had so far.

> Alternatively, we could spend a bit more time thinking about what sorts
> of documentation a Wiki would be good for.  This might give us a better
> handle on the critical mass question.