Re: Using network shares as sandboxes

[Kaz Kylheku]

In article <address@hidden>, brent wrote:
>I have been informed that there are 'a thousand reasons' why
>not to use network shares as sandboxes with CVS.

You should not use a remote filesystem as your CVS repository. 
Use CVS in client/server mode instead. The CVS software uses the
filesystem as a database, relying on certain atomicity properties
of the various filesystem operations. If the networked file system
does not guarantee the properties in all situations, that can
spell trouble for the integrity of the repository.

There is less of a clearcut case against using a remote filesystem
for a working copy. That is what a ``sandbox'' is.

>sandboxes with no difficulties.  Presently I wish to use CVSNT 
>as the repository server with Win2K shared drives as sandboxes.  
>Mostly, these sandboxes will be accessed from Win2K machines, 
>but we also have Linux machines wishing to use the same sandbox.  
>Only one user would be actively using the sandbox (ie issuing a 
>CVS command), but we wish to keep things on the network to make 
>it easy to communicate and share files between team members.

This is nonsense. Each machine should have its own checked out copy.
The differences between representations of text files are going to
cause problems.

CVS supports sharing at the repository level. There is no provision
for sharing the same working copy among users or machines.  It's 
not supported by the data representation of the working copies,
or the concurrency model.