Re: 2.1 Fallout; httpd v.s. httpd-2.0

[Aaron Bannert]

On Monday, November 25, 2002, at 12:22  AM, Wilfredo Sanchez wrote:

On Sunday, November 24, 2002, at 11:48  PM, Aaron Bannert wrote:

We can make a duplicate of the httpd-2.0 CVS module and call it
httpd-2.1 or whatever the heck we want, and keep the history. Why do
we have to lose the history?

-1 to losing the history

This loses future history. That is, ongoing work on 2.0 becomes detached from the 2.1 repository, and you may as well just start a new module if going back to the old one in necessary anyway.

This was merely a possible solution to Justin's statement that we would *have to*
lose history if we created another new CVS module.

I'm still +1 for branches.

I don't really buy this performance noise about branches. In Darwin CVS, the Core OS team make a new branch *for every bug fixed* (my fault it's done that way). Each branch is merged down individually as the changes are approved by the group, and is allows engineers to hack away and commit as they go without disrupting anyone else. I'm talking about several hundred branches, and it's all running fine on some G4 box in a lab.

As Mark said, long-lived branches do take a hit, but that's why you branch off the maintenance release on leave active development on > HEAD.

So are you suggesting that it might be better if we branch off smaller branches
for each major.minor release instead of just for each major release?

eg. For now we'll have 2.0 branch, 2.1 branch, 2.2 branch...
Would you propose 2.0.44 2.0.45 2.1.0 2.1.1 2.0.46 2.1.2 ... ?

I'm not entirely sure how that would work...

I do not fear the branch. The branch is puny and I will crush it with my pinky.

Um, same here... :)

-aaron