Re: using cvschroot allowed?

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Harald Dunkel <harald@CoWare.com> writes:

> Mark D. Baushke wrote:
> > Harald Dunkel <harald@CoWare.com> writes:
> > 
> > 
> >>> Question: Is it allowed to use cvschroot in this way? Or is
> >>> this too much of a hack, maybe corrupting my CVS repository
> >>> at checkin? Anything I should take into account to make this
> >>> work?
> > 
> > Well, I would try to avoid excessive use of this technique as an
> > interruption during the altering of the CVS/Root process might leave
> > your tree in an odd state.
> > 
> 
> I see. So either it should be made very hard to interrupt cvschroot,
> or I should explicitly use 'cvs -d $CVSROOT_WRITEABLE commit', etc.
> instead, since setting $CVSROOT is not sufficient to override CVS/Root.

Yes.

> > Please note that there are two separate methods available to have a
> > read-only CVS repository that may be used for checkout and without any
> > changes to the user's workspace.
> > 
> >   1) Allow a commit via a writeproxy configuration (needs cvs 1.12.x
> >      where x > 9). Note: Not many folks are using this feature right
> >      now.
> > 
> 
> Its been some time since I looked at this feature, but AFAIR it did
> not support pserver, and the documentation about setting the
> scripting hooks and the necessary file system locks was a little
> bit "thin".

True.

It also assumes that ssh-agent forwarding is happening to the CVS
secondary server in order to allow it to connect to the primary CVS
server in write-thru mode which is a potential security problem. Letting
the clients do the redirct is usually better. Generally, you only need
the write-thru mode when you have an older client. If you have a new
client, then the redirect to the primary server is only slower, but more
reliable than if you had added a cvs -d :ext:primary/path/to/root'
switch to your cvs command lines that do write operations.

> >   2) Use a commercial product add-on to CVS from WANdisco.com
> > 
> 
> Of course there are many commercial version control systems available,
> but lets assume that this is not an option.

The only reason I mention it is that they use :pserver: for their
implementation, so it works fine in the multisite environment that is
not otherwise supported with CVS right now.

> > You may find doing a goole for 'cvs multisite' and 'cvs writeproxy' may
> > be useful to you.
> > 
> 
> I will check. Many thanx

Good luck.

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.4 (FreeBSD)

iD8DBQFE7J6ZCg7APGsDnFERAvaGAKDfR39CX+jqqI3yOAMpmtUxo/Nt5gCgz7uH
8QZo+MplSDVvQ0Au00nbkFc=
=5jy7
-----END PGP SIGNATURE-----