Lars Gullik Bjønnes wrote:
Abdelrazak Younes <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes wrote:
| > Abdelrazak Younes <[EMAIL PROTECTED]> writes:
| > | > then point us either at the patch or the revision number.
| > | > Then patches can (as long as they are not dependent on eacy other) be
| > | > cherry picked and merged easily to trunk.
| > | | Talk about an easy procedure. It's too complicated for me, so no
| > thanks.
| > You are of course aware that this is (basically) how you would work
| > with a distributed scm?
| > It might be a tiny bit more cumbersome with svn, but no real
| > differences.
|
| Nope there is a big difference. You are basically asking me to have 3
| trees.
Actually I am asking you to have as many trees as necessary.
(branches really...)
Locally they are trees. Given that each tree is 830 Megs I have enough
of 3. My head is not scalable sorry.
| One for trunk, one for small changes to trunk and another one
| for my bigger development.
And perhaps even more.
(unicode, any, pristine-trunk, work-trunk, younes and a couple of more
is the trees I have checked out currently.)
Actually I have tree: pristine-trunk, work-trunk and younes.
| Why a distributed scm, everything is local
| and branches are real branches not cheap copies.
What is the difference between a branch and a copy? (and when is a
copy cheap?)
I won't expand because here I don't know enough about that stuff. I am
just repeating the arguments that I read elsewhere.
| At least that's what
| I understand reading git user manual.
(can you point me at the git manual please?)
But with git you still have to ask other to pull (merge a revision)
And even in your local tree with git (or others) you would have more
than one branch for different work?
The main difference I see is that you can be completely off-line, and
do not have to syncronize with anyone else when working in your git
tree. Otoh it is just has hard to get others to accept your changes or
to move them between git trees (barring syntax differences).
From what I've read, the merging algorithms of git are much more
powerful than those of svn. But that is not really my point. Working
with SVN means either that one guy controls everything (you) or that
each developer is trusted to do the right thing (comments are of course
welcome and encouraged). Sure this is achievable with SVN but this is
de-facto like this with a decentralized scm because anyone owns its own
repository. The other obvious advantage is that you can work off-line.
(Sure svn has a harsher requirement in that meging must be done
between branches from the same repository.)
Yes.
Abdel.
