[prev in list] [next in list] [prev in thread] [next in thread] 

List:       dcms-dev
Subject:    Re: [opencm-dev] About Running OpenCM though SSH
From:       Ethan Blanton <eblanton () cs ! ohiou ! edu>
Date:       2003-12-04 17:20:56
[Download RAW message or body]

Jonathan S. Shapiro spake unto us the following wisdom:
> Actually, I've been thinking more about this. I'm still not entirely
> convinced that it's a good idea, but it shouldn't be hard to build a
> modified server that would run on stdin to handle this case.

I have a couple of points in the "why this is a good idea" camp:

* I have some repositories (currently CVS for most things, but all new
  projects have been going into OpenCM recently) which contain things
  which only _I_ care about, and may care about only on a particular
  machine; dotfiles, one-off projects, personal scripts -- I am not
  super excited about running an OpenCM server on every machine on
  which I wish to have such a repository, but on the other hand
  conflating them (across work domains, schools, etc.) does not really
  make sense either; to further complicate things, some machines which
  do not share filesystems may share such a repository in the perfect
  world.  CVS's ssh-based access is perfect for this.

* While http is of course the king of this phenomenon, there are many
  places which I can get ssh into and out of, but from which I cannot
  necessarily access a random (for instance, 3434) port.  I currently
  use ssh tunnels for this and connect to a proper OpenCM server; this
  is suboptimal for an opencm repository which really has no other
  users.

* Very secondarily, this also opens the possibility for some
  cooperation with inetd with little effort -- I periodically restart
  the OpenCM server on my box (when it gets above 7% CPU usage doing
  nothing (I assume actually garbage collecting?)), but it would be
  equally handy if xinetd simply spawned it when I needed it and it
  died after some period of inactivity.  As my usage patterns are
  bursty, this would be effective without incurring a painful
  overhead.  The only additional modification such a setup would
  require (that I see, there may be more) would be an option for the
  server to shut down after some period of inactivity.

> In the meantime, I am currently using a shell script to accomplish much
> of what I suspect Hilary is really trying to get done.

Aside from the third (and, as I said, secondary to me) point, my
personal needs could be served by a relatively trivial script; in
fact, I have written scripts to do some things.  It seems to me that
it is important for something like the CVS ssh transport to be
available "out of the box", but providing it via a script gives me no
heartburn at all...

Ethan

--=20
To surrender one's personal weapon is to invite disaster. This has
been obvious for so long and so often that there is probably a Greek
word for the practice.
		-- Jeff Cooper

[Attachment #3 (application/pgp-signature)]
_______________________________________________
opencm-dev mailing list
opencm-dev@smtp.opencm.org
http://www.opencm.org/mailman/listinfo/opencm-dev

[prev in list] [next in list] [prev in thread] [next in thread]