[cvsnt] Re: virtual branches ?

[Oliver Giesen]

Tony Hoyle wrote:

> Prochazka, Jan wrote:
> > Wouldn't be possible to specify at the beginning at revision 1.10 a
> > "virtual branch" that would become real branch if and only if
> > somebody commits to the branch ? Until that, any update of branch B
> > returns HEAD of file F, because the real branch not exist yet.
> 
> Yes, that's called a floating or 'magic' branch (tag -M).  They're
> not as useful as they seem in practice, but the option is there.

Cool, I wasn't aware you managed to fix these. I remember trying them
quite some time ago and they weren't really working at all and that was
the last I heard or read about them since. Now after reading this I
gave them another try and all went fine! :)

Well, almost...

With one file I got the following problem:

cvs tag -M Magic-BRANCH test.txt
[modify test.txt]
cvs ci test.txt
"cvs [server aborted]: Revision 2.9 is already locked by HQ\GiesenO"

Could the problem be the forced revision number here? At least that's
the only difference I was able to make out in comparison to the files
where it worked. I definitely did not lock the file in any sense I
could think of (i.e. admin -l, edit -x, -kx). Access was via :sspi: to
a local 2115 server (using the same as client) and HQ\GiesenO is my own
user name.


BTW: What would you say would be the minimal server version requirement
for using magic branches, i.e. since when are they "stable" (or at
least since when are they in their current state)?

Also, do you think there's any way to turn an existing branch into a
magic one? What about the other way? Can I make it stop floating
without having to force commit a (dummy) branch revision?

Cheers,

-- 
Oliver
----  ------------------
JID:  ogiesen at jabber.org
ICQ:  18777742	(http://wwp.icq.com/18777742)