Equivalent to svn tags?

Matthew D. Fuller fullermd at over-yonder.net
Thu Oct 25 13:29:14 BST 2007


On Thu, Oct 25, 2007 at 08:06:21AM -0400 I heard the voice of
Aaron Bentley, and lo! it spake thus:
> 
> Have you seen bzrtools' multi-pull?

Yes, but it's a solution to a somewhat different problem, which has
some overlap with the one I was aiming at, but isn't the same thing.
As a few quick differences:

- It's a pull, not a branch, so I still have to individually branch
  all the branches.  Closely related:

- It doesn't bring in NEW branches.  Any time a new branch appears, I
  have to somehow get notified and manually create it.  That's big fat
  suckage.

- It pulls "all branches at this location in my fs", not "all branches
  in this logical set", which can have other side effects.  For
  instance, I may have my own branches around there, that have one of
  the upstream branches as parents, and I may end up pulling over a
  currently-fully-merged branch where I didn't want that mainline
  trashed.

- It treats it all branch-at-a-time still.  Doing a no-op pull can
  easily take several seconds just to find that out.  In a
  multi-branch setup, they'd all be coming from the same place, so all
  of them could be checked at once, and 5 seconds later I'd see
  "nothing to do".  If it takes 5 seconds to find out there's nothing
  to do for each branch, though, and I have 15-20 sitting there...  it
  takes minutes just to do nothing (this is another SS topic
  intersection).



This is one of the reasons I dislike these discussions; I always feel
like I end up attacking, or other people think I'm attacking, what may
be perfectly good solutions to overlapping-but-different problems.  I
don't mean it that way, and I don't intend to denigrate the
extraordinary effort put in by a lot of people into making incredibly
useful solutions to a huge range of problems.

It's the same thing that always happens to me in PQM discussions; it
may be a solution, it may be a good solution, it may be a good
solution to a related problem, but it's not quite the _same_ problem,
so no matter how good it gets at solving its problem, it's probably
never going to be a good fit to the little monster I'm facing.

There are real obstacles to getting good fit solutions to all (or even
all the major) problems.  But I don't think they're all technical, or
inherent in the positions we've staked out bzr to hold.  The biggest
obstacle is not "we can't do that and have bzr remain bzr", it's
cycles; it's "we can put that on our todo list right after the higher
priority stuff; how's 2015 for ya?"


-- 
Matthew Fuller     (MF4839)   |  fullermd at over-yonder.net
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.



More information about the bazaar mailing list