[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

read-only checkout when developer has no commit access

From: Jason Elliot Robbins <jrobbins_at_collab.net>
Date: 2000-07-25 01:45:48 CEST

This started as part of a discussion on another tigris list.

The issue is that users who checkout files as guest get read-only
copies. That is a problem because often one moust modify Makefiles or
config files.

>[readonly files out of readonly repos]
>On Mon, Jul 24, 2000 at 03:14:59PM -0700, Niels P. Mayer wrote:
>> Again, it's behavior i've seen in CVS prior to tigris. Perhaps it's a
>> config option i don't know about.
>As long as I've used CVS (with mostly default options), all my
>checkouts from any anoncvs server but Tigris's have resulted in
>writable files.
>> I'm still inclined to leave the current behavior, which is semantically
>> correct
>I disagree, for reasons others have given.

This is a feature of tigris and cvs. Apparently, the pserver protocol
allows the server to tell the client whether each file should be ro or
rw. The idea is clearly to provide a reminder to the developer as to
what commit access they will have before they go down a long road of
modifications that cannot be checked in. There is support for
controling this feature in the client, but apparently not enough or it
it is broken, because -w did not work for me, and WinCVS has no option
to make files rw if they are not sent from the server that way.

It seems like a mis-feauture to me. I would rather let developers
hack into all files and then give them help making a patch file if
they cannot commit.

In fact, this is part of the motivation for requesting local
repository support in subversion. I sould like my version control
system to make files rw, and let the developer commit them to the
local repository, and when they are ready to upload to the main
repository, let anyone do it on their own private branch.

As repository administrator on an open source project, I would rather
tightly control permission to promote/merge int the main branch than
control premission to commit. I want per-branch permissions in
general. There would also be a need to manage developer disk space
quotas gracefully, but that is way down the road.


Jason Robbins, Ph.D.      Collab.Net is hiring open source developers!
Chief Architect                             http://www.collab.net/jobs
Received on Sat Oct 21 14:36:05 2006

This is an archived mail posted to the Subversion Dev mailing list.