(I'm de-lurking from under this snow cover for a second)
On Dec 17, 13:40, Martin Buchholz wrote:
>It would be better to use the work `branch' in the name release-21-2,
>to make it clearer what is a branch and what isn't. This distinction
>is hard to understand until you study it for a while.
With the naming convention used it can indeed be confusing and it would
probably help with having `branch' in the name. Unfortunately it isn't
really possible to change the branch name after the fact :-(
(there is an aliasing function in CVS (or rather, cvs admin) but it's
broken as far as I can tell and has been for a while).
>ST> Or maybe retire "release-21.2" and use "r21-2-latest-beta" (or one of
>ST> the two obvious abbreviations, I favor "r21-2-beta" for brevity and
>ST> clarity) for the most recent consistent sources and "r21-2-head",
>ST> "r21-2-current" (I don't like it, too similar to "latest"), or
>ST> "r21-2-bleeding-edge" for the possibly inconsistent head of the
>ST> branch?
>
>Again, the problem is that the head of the branch itself doesn't have
>a name, but the branch as a whole does. And then there's the trunk -
>it doesn't even have a name :(
The branch name *is* effectively the same as the head of the branch itself.
To get the (possibly inconsistent) head of the branch you just do
'cvs update' if you checked out the branch, or 'cvs update -r branchname'
if you didn't. And you can do things like
cvs diff [-r whatever] -r release-21-2
to compare your current files [or optionally version 'whatever'] with the
latest-and-greatest on branch 'release-21-2'.
So there shouldn't be any need to tag the latest-and-greatest on a branch,
it's probably not even a good idea.
It is indeed painful that the CVS trunk is nameless. If you don't tag it
there is no way to refer to the head (as -rHEAD doesn't really work that
way), the only thing you can do is to check out the head with update -A.
So you can get to the files but you can't refer to them with 'cvs diff'
etc., at least not in a reliable way).
>ST> Sorry, I should have thought about this before when you posted your
>ST> "Engineering Changes" proposal. But it just hit me smack in the face
>ST> when you juxtaposed a tag named "release" with "may not even compile"!
>
>Hmmmmm...
>
>dev-21-2-branch
>
>branch-21-2
>
>dev-21-2
Those would indeed be better names than the release-21-2 name..
>Martin
-Tor
