Hello.
[...]
> Justin Erenkrantz <jerenkrantz@apache.org> writes:
> > Please remember that CVS respects permissions - it's an inherent flaw
> > in SVN's model that it can't. A working copy in CVS has at least
> > the same permissions as its ,v file. So, for true CVS compliance, you
> > need to implement something like this.
> >
> > I doubt I'm the only one who has relied on the permissions feature
> > of CVS. So, here's a case where SVN 1.0 will not be as good as CVS
> > if this feature isn't present. -- justin
>
> How do such CVS working copies behave under Windows, or other non-Unix
> OS's?
Don't know.
> And what does it mean when the same users and groups don't exist on
> one working copy box as on another?
AFAIK, this is only about permissions, but not about user/groups,
i.e. the rwx flags for "ugo" are preserved. The user and group are
simply set to whoever you happen to be at the moment of checkout.
> Can you describe exactly how you depended on this feature of CVS,
> maybe?
I also depend implicitly on this feature... without having thought
much about it until now. It may not be normal (i.e. intended) use, but
out of some need, I happened several times to use a working copy
directly for production use, i.e. without an install procedure to copy
the data into its final location, but checking it out to its final
location directly. Not pretty, but it works.
Regards,
Benjamin.
- application/pgp-signature attachment: stored
Received on Thu Aug 22 13:32:23 2002