On Jul 4, 2006, at 4:21 AM, Aaron Mulder wrote:
Are you recommending that the project switch to Perforce, or is this
just wishful thinking?
Well... both I guess. But given our reality its more of the later.
Also, one more note on the merging -- If we change the directory
layout in the branch, it will be very hard to merge any patches to
HEAD into the branch to keep them in sync since all the patch paths
will be wrong. I don't see how this can end well. Unless there's a
freeze in HEAD for the duration of the branch, which would mean we're
just using the branch to work around RTC.
The reason why I mention it over and over... is that Perforce can
handle automating a bunch of this work.
For example, to solve the issue you mention about patches. If you
know the change number that the patch was done against, then you
could create a new temporary branch based on that change number, then
apply the patch, and then merge the private branch back into the
trunk. Perforce is smart enough to realize what needs to merge and
what has been merged already since it tracks integrations. It also
knows what been moved and deleted and when you merge the changes will
follow the code wherever it was moved to.
It really is an excellent tool... for highly branched and dynamic
environments Perforce really shines.
--jason
Thanks,
Aaron
On 7/4/06, Jason Dillon <[EMAIL PROTECTED]> wrote:
FYI... Perforce may be commercial, but they provide free licenses for
qualifying open source usage:
http://perforce.com/perforce/opensource-faq.html
:-)
--jason
On Jul 4, 2006, at 3:50 AM, Aaron Mulder wrote:
> Jacek,
>
> When discussing whether a branch was appropriate, I expressed a
> concern that it would be difficult to merge changes from this
branch
> to HEAD because SVN seems to have difficulty handling multiple
> revisions of add/delete/move/copy operations in a single merge
(and I
> understand the M2 restructuring will involve a lot of that). I've
> often had problems where something is e.g. removed and then
recreated
> and I try to apply and it refuses claiming that there's
something in
> the way. The only way to proceed is to manually delete offending
> directories and then update. But that won't work for merging
from a
> branch to HEAD -- it would effectively be a revolution operation
where
> we would have to vote to delete HEAD and move the branch to HEAD.
> Then we lose the bug fixes that have been applied to HEAD and we've
> been through that before.
>
> I'm concerned that you ignored this concern and went ahead with the
> branch plan. In fact the only response I got was that perhaps a
> commercial source control system would be better. Perhaps that
means
> this should have been an RTC operation so you were forced to
address
> my concern before taking this approach.
>
> Thanks,
> Aaron
>
> On 7/4/06, Jacek Laskowski <[EMAIL PROTECTED]> wrote:
>> Hi,
>>
>> Just created a branch - m2migration - for all our work
pertaining to
>> the M2 migration of Geronimo build. Everybody's welcome to work
on it
>> *without* RTC on. Revolutionary rules are enabled again! ;-)
>>
>> The branch is available at
>> https://svn.apache.org/repos/asf/geronimo/branches/m2migration.
>>
>> Once we're ready to move a part of the work done in this branch,
>> we'll
>> create a patch and RTC'ed it to commit to trunk. I believe it will
>> help those who are reluctant to work on the branch and give a hope
>> not
>> all is/will be lost ;-)
>>
>> Just to be clear: anybody who wants to learn M2 tricks and help us
>> with the migration is welcome. Those who aren't committers can/
should
>> count on me to commit their work (my kids are away on their
vacation,
>> so let's do it before they come back home ;-))
>>
>> Jacek
>>
>> --
>> Jacek Laskowski
>> http://www.laskowski.net.pl
>>
