3 messages in com.perforce.perforce-userBranch question| From | Sent On | Attachments |
|---|---|---|
| Rich...@netapp.com | 25 Jun 1999 15:21 | |
| Stev...@Eng.Sun.COM | 25 Jun 1999 16:16 | |
| Rick...@vsl.com | 25 Jun 1999 16:24 |
| Subject: | Branch question![]() |
|---|---|
| From: | Rick...@vsl.com (Rick...@vsl.com) |
| Date: | 06/25/1999 04:24:05 PM |
| List: | com.perforce.perforce-user |
On Fri, 25 Jun 1999, Steve Cogorno wrote:
I have a question about the internal representation of a branch. When I branch for a release, the majority of files don't change (about 75%). Does perforce simply reference the version in the main branch if there were no changes?
Yes, I see it as being similar to "Copy On Write" memory pages for shared libraries, if you're familiar with that.
An interesting thing that I saw when looking at this a few days ago is that when you integrate file changes from a branch back to the mainline where the file came from, the mainline file doesn't physically change. Perforce seems to keep the sequence of revisions elsewhere and assembles them as needed. If you obliterate the branch file after this integration, _then_ the mainline file changes. It splices in the changes that were stored in the branch in order to delete the branch file.
In this same scenario, I also noticed that the change comments from the branch, which where visible before the obliterate with "p4 changes -i -l", seem to be gone after obliterating the branch. The file changes themselves, as I mention above, are now stored in the mainline file because of the integrate and of course not lost.
It seems like an oversight or a bug not to keep those branch revision comments since the branch revisions have been integrated into the mainline. Just yesterday I asked Perforce about this, to see if it a bug, but I haven't heard back yet.
After the obliterate, the only change comment that remains in the one for the integrate from the branch to the mainline.
Is this how others see and understand the internal operation?
...RickM...




