>>>> "SJT" == Stephen J Turnbull
<turnbull(a)sk.tsukuba.ac.jp> writes:
>>>> "vin" == Vin Shelton <acs(a)xemacs.org>
writes:
vin> As long as I have a tag to use, I don't care when I get moved
vin> off the main. Unless there's some subtlety I'm missing.....
SJT> Not that I know of. But getting the tag in place correctly will
SJT> require that you finish up whatever you're doing at the moment, then
SJT> cvs tag release-21-1, then update -r release-21-1, then start applying
SJT> stuff again. AFAICT, _you_ can do this anytime you want and just
SJT> announce that people have to use the tag from now on.
You have to do more than do a plain tag, you have to (also) make a branch tag.
SJT> But if somebody else does it, I could easily see somebody else going
SJT> "oh, yeah, I was gonna move the HEAD today," cvs tag, cvs do lots of
SJT> stuff to move HEAD, right in the middle of you doing a bunch of
SJT> pre-release commits. Eek.
SJT> The other thing is that the last time there was serious discussion of
SJT> moving HEAD, somebody who claimed to have done it said we could expect
SJT> to have to take CVS down for a day or so. I really think the
SJT> Honorable Stable Release Manager might have a thing or two to say
SJT> about _that_. :-)
I don't like to restrain developers. So for example, currently there
are no scheduled CVS commit freezes for beta releases. However,
moving the CVS branch will quite likely mean a day of no-commits and
perhaps many more days of preparation by the cvs administrator.
SJT> Also, Tor doesn't say, but I wonder what happens to CVS history data
SJT> when you do this.
Depends on how you do this. But expect to lose the coherence of CVS
history, unless deep magic is invoked. I mean deeper magic than the
relatively straightforward method Tor describes.