I'm running p4 interchanges -b my_branch, and I get a ton of results, the first one being a changelist that we integrated a long time ago.
So I try to integrate again, but p4 integrate -b my_branch //...#changelist,#changelist just returns "All revision(s) already integrated".
The only way to unblock this is to do a forced integration (-f in the integrate command) and then simply accept target (-at when resolving), and that works - p4 interchanges then no longer lists this changelist.
But how can Perforce get into this state to begin with? This happened after we've done a bunch of integrating across multiple branches, but I nothing that I'd think would cause a changelist to become "unintegrated" somehow.
This is on a 2014.1 server.
Thank you for specifying your server version.
The 'p4 interchanges' command can give the "All revision(s) already integrated" message with misleading results when cherry-picking is involved.
There is a command line example here:
http://answers.perforce.com/articles/KB_Article/Cherry-Picking-Integrations
You could also be affected by a bug that was patched in 2014.1 listed here in the server release notes:
http://www.perforce.com/perforce/doc.current/user/relnotes.txt
Bugs fixed in 2014.1 PATCH5
#880506 (Bug #71725) **
The istat.mimic.ichanges configurable controls the reporting
of revisions between stream and parent. If set, istat will
not report cherry-picked revisions already present in the target.
The default behavior will report any changes not credited, even
when the content may already be in the target.
If you would like, you can pull the most recent build of the server P4D for your OS from our ftp site: http://ftp.perforce.com/perforce/r14.1/
REFERENCE
http://answers.perforce.com/articles/KB_Article/Integration-Changes-Reporting
Related
As mentioned in the question, I often encounter the situation shown in the picture below:
Problems demonstrate
I'm using the Clean... directive here, not the Reconcile Offline Work...directive
Clean...
This bug was fixed in the server in release 2016.2:
https://www.perforce.com/perforce/doc.current/user/relnotes.txt
#1382996 (Job #74886, #86396) **
'p4 clean' would fail to sync files when needed when they
are exclusively opened by another client. This has been
fixed.
Double check that you're on release 2016.2+ of the server; if you are, you should be at the very least be able to use p4 clean as a workaround (e.g. by adding it as a custom tool to P4V).
I am facing a weird behavior in perforce submitted files, after p4 integrate
Scenario
Repo contains C# code and the built DLL. I wished to integrate a changeset from one branch to another
so I am following the steps which I had discussed, a while ago
The .cs files doesn't complain. If there is resolve conflict in the DLL, I choose either of Accept Source or Accept Target without much care. Reason being, I always rebuild the DLL before submitting
Issue
The DLL which was built in local workspace and reflected in the changeset is correct. I tested it locally and do p4 submit. But to my amazement, the DLL submitted is not the one which I had built. Instead the one which came from that other branch was submitted
Confusion
I thought that with perforce, when a file is opened in a changeset, always the latest(local) copy will be submitted. That is why I don't pay much attention if there are conflict reported in DLL
Isn't that correct ?
Why would the submitted file be different from my workspace version ?
When you "accept source" you're recording that you want the target file to be an exact copy of the source file; consequently, if you submit the file, it's not even transferred from the workspace (in order to save time) -- instead it's just copied server-side. If you tampered with the workspace file, this leads to the situation you describe where the workspace is now inconsistent with the depot, just like if you'd modified a file that wasn't open for edit.
If you submit with the -t flag (for "tamper checking"), it will check for tampered files by comparing what's in your workspace with what should be in your workspace according to the resolve options you picked:
C:\Perforce\test\integ>p4 integ source target
//depot/integ/target#2 - integrate from //depot/integ/source#3
C:\Perforce\test\integ>p4 resolve -at
c:\Perforce\test\integ\target - vs //depot/integ/source#3
//Samwise-dvcs-1509687817/integ/target - copy from //depot/integ/source
C:\Perforce\test\integ>echo tampertampertamper >> target
C:\Perforce\test\integ>p4 submit -t -d "submitting tampered file"
Submitting change 190.
Locking 1 files ...
integrate //depot/integ/target#3
//Samwise-dvcs-1509687817/integ/target tampered with after resolve - edit or revert.
Submit aborted -- fix problems then use 'p4 submit -c 190'.
Some file(s) could not be transferred from client.
If you p4 edit the file, it's changed from a pure copy to an edit, and will be read from the workspace instead of from the source file:
C:\Perforce\test\integ>p4 edit target
//depot/integ/target#2 - reopened for edit
C:\Perforce\test\integ>p4 submit -c 190
Submitting change 190.
edit //depot/integ/target#3
Change 190 submitted.
In an analysis of the history in our perforce depot, I need to identify those changes where a full integration from another branch was performed and also find out the exact change number of the source of that integration. Unfortunately, even though such a full integration is a common operation, I could not find any easy and reliable way to detect it in the history. Please let me know if I missed something.
In any case: via a set of scripts using 'p4 filelog' and matching up revision numbers, I managed to find all candidate changes and their respective integration source change. What I am missing is a means to distinguish full integrations from cherry-picks or partial integrations limited to a subdirectory. For this, the closest I could find is the 'p4 interchanges' command, which does exactly the thing I need, except for the problem that the 'toFile' argument cannot have an '#' revision specification.
I would have hoped that
p4 interchanges //depot/sourcebranch#123400 //depot/targetbranch#123499
would tell me whether any changes were missing in the integration point I found, but it only gives the error 'A revision specification (# or #) cannot be used here.' - which matches the documentation.
Is there any other means to examine integration points in the p4 history to distinguish cherry-picks from full merges?
Use the undoc p4 integ -C changelist command:
p4 integrate -1 -2 -C changelist# -Rlou -Znnn
... The -C
changelist# flag considers only integration history from changelists
at or below the given number, allowing you to ignore credit from
subsequent integrations. ...
Hence:
p4 integ -n -C 123499 //depot/sourcebranch/...#123400 //depot/targetbranch/...
should tell you whether sourcebranch#123400 was fully integrated into targetbranch#123499. If you use -C 123498 in theory the difference in the output will show you which files were integrated.
There are probably some edge cases around deleted files -- e.g. if you integrate a deleted file into a file that's deleted at the head revision, it will report "up to date" regardless of integration history, so I can imagine that a file that was skipped for that reason but then subsequently re-added might give a false positive with the above method. (Or it might not -- I have vague memories of possibly fixing that scenario, but undoc bug fixes don't show up in the relnotes...)
Here's an example where foo#2 was integrated into bar#3:
sams-mbp:test samwise$ p4 filelog ...
//stream/main/test/bar
... #2 change 4 delete on 2019/07/21 by samwise#samwise-dvcs-1517552832 (text) 'delete'
... #1 change 3 branch on 2019/07/21 by samwise#samwise-dvcs-1517552832 (text) 'branch'
... ... branch from //stream/main/test/foo#1
//stream/main/test/foo
... #2 change 6 edit on 2019/07/21 by samwise#samwise-dvcs-1517552832 (text) 'edit'
... #1 change 2 add on 2019/07/21 by samwise#samwise-dvcs-1517552832 (text) 'add'
... ... branch into //stream/main/test/bar#1
sams-mbp:test samwise$ p4 integ -n -C 2 foo#2 bar
//stream/main/test/bar#2 - branch/sync from //stream/main/test/foo#1
sams-mbp:test samwise$ p4 integ -n -C 3 foo#2 bar
foo#2 - all revision(s) already integrated.
With -C 2 (before the branch), we see a replay of the integration as it would have happened as of that point in time. With -C 3 (the changelist of the branch), we see "all revisions integrated" because by that point in time it had already happened.
After continuing to try out all combinations of arguments, I finally found a solution that works for me:
p4 interchanges -b SOURCE_to_TARGET //depot/targetbranch/...#targetchange
will print out all changes on the source branch that are missing on targetbranch#targetchange. It will return only changes older than targetchange. If this list contains no changes older than sourcechange, the targetchange was a full integration.
The command may take considerable time to complete when the returned list is long. Unfortunately, I can't find a way to truncate this search, but I can live with that.
As it seems, some of this functionality was buggy up to server version 2018.2 which might have caused difficulties in my earlier attempts.
I am trying to integrate a branch with several files that were moved. They were not modified in target branch, yet appear as conflicts. When clicking "accept source" file silently disappears from conflicts dialog, but still remains marked as conflict in the changelist. One way I can fix this is adding -Di flag, but it's a big no-no on my team. I also tried p4 resolve through command line, with same outcome (no error or anything, but nothing is resolved).
What is going on and how can I resolve preserving the move history?
Do:
p4 resolve -as
from the command line. In the vast majority of cases that's all you need to do; it tells resolve to accept the change relative to the base (which means accepting the source in the case where the target hasn't changed).
(updating to include additional info from comment)
If after you do that you get an error like:
can't move (open for delete); must accept other resolve(s) or ignore
it means that the file was moved in the source (which means that normally a resolve -as would move the workspace file to match it), but the file in your workspace can't be moved because it's already open for delete (you can't move a deleted file). This is a pretty rare situation that happens if you move a file, delete it, and then try to resolve those two operations independently (but without submitting in between). In that case you probably want to specifically "ignore" the move resolve (like the error message suggests) by doing:
p4 resolve -ay
If you've somehow gotten your working file into a bad state (maybe you were running random commands before setting up the integrate and the workspace wasn't in a clean state) and either can't figure out how you got here or have no interest in doing forensics when you just want to do a basic integrate and forget about whatever you were in the process of doing before, you can always start over like this:
p4 revert FILE
p4 integrate -b BRANCH FILE
p4 resolve -as
If you have a file that was moved and then deleted in the source, it's not possible for both of those actions to be represented in a single changelist in the target, so the default is to ignore the move and accept the delete (on the theory that since the file is deleted anyway, it doesn't particularly matter where it's deleted):
C:\Perforce\test\movedel>p4 integ A/... B/...
//stream/main/movedel/B/foo#1 - integrate from //stream/main/movedel/A/bar#1,#2 (remapped from //stream/main/movedel/B/bar)
C:\Perforce\test\movedel>p4 resolve -as
c:\Perforce\test\movedel\B\foo - resolving move to //stream/main/movedel/B/bar
//Samwise-dvcs-1509687817/movedel/B/foo - ignored //stream/main/movedel/B/bar
c:\Perforce\test\movedel\B\foo - resolving delete from //stream/main/movedel/A/bar#1,#2
//Samwise-dvcs-1509687817/movedel/B/foo - delete from //stream/main/movedel/A/bar
Note that after resolving move we see an ignored and after resolving delete we see delete. Note also that the history of B/foo is linked to A/bar, which contains the history of the move, so the history isn't "lost", it's just not duplicated.
If you do want to duplicate the move history from the source into the target (so that the target, viewed independently of the source, shows the file as having been moved and then deleted), you need to do the integration in multiple submits:
C:\Perforce\test\movedel>p4 revert ...
//stream/main/movedel/B/foo#1 - was delete, reverted
C:\Perforce\test\movedel>p4 changes A/...
Change 316 on 2022/12/08 by Samwise#Samwise-dvcs-1509687817 'delete bar'
Change 315 on 2022/12/08 by Samwise#Samwise-dvcs-1509687817 'move foo to bar'
Change 313 on 2022/12/08 by Samwise#Samwise-dvcs-1509687817 'add foo'
C:\Perforce\test\movedel>p4 integ A/...#315 B/...
//stream/main/movedel/B/foo#1 - integrate from //stream/main/movedel/A/bar#1 (remapped from //stream/main/movedel/B/bar)
C:\Perforce\test\movedel>p4 resolve -as
c:\Perforce\test\movedel\B\foo - merging //stream/main/movedel/A/bar#1
Diff chunks: 0 yours + 0 theirs + 0 both + 0 conflicting
//Samwise-dvcs-1509687817/movedel/B/foo - copy from //stream/main/movedel/A/bar
c:\Perforce\test\movedel\B\foo - resolving move to //stream/main/movedel/B/bar
//stream/main/movedel/B/bar - moved from //stream/main/movedel/B/foo
C:\Perforce\test\movedel>p4 submit -d "integrate #315"
Submitting change 317.
Locking 2 files ...
move/add //stream/main/movedel/B/bar#1
move/delete //stream/main/movedel/B/foo#2
Change 317 submitted.
C:\Perforce\test\movedel>p4 integ A/... B/...
//stream/main/movedel/B/bar#1 - delete from //stream/main/movedel/A/bar#2
C:\Perforce\test\movedel>p4 submit -d "integrate #head"
Submitting change 318.
Locking 1 files ...
delete //stream/main/movedel/B/bar#2
Change 318 submitted.
What is going on
Perforce is not able to deal with moved files in merge/integrate properly. I can confirm that this is still an issue in the most recent version. We have this case regularly.
Say, team A is working on //OurDepot/ExclusiveContentOfTeamA/ and you are then merging that to team B's branch. Consider the case where team B has not touched (or even looked at) that location.
Merging should be a breeze in this scenario, right? Wrong!
If team A moved and deleted files, it is always a dice roll whether the merge will work or not. Most of the time "accept source" will resolve the issues, but sometimes you get the state you described in your question.
and how can I resolve preserving the move history?
You can't. All you can do is some hacky solution forcing perforce to delete the files (p4 resolve -ay), and then manually repair the state in the branch by re-adding the files as new files. History will be lost.
I am using following command to sync B vob files from A vob
clearfsimport -master -follow -nsetevent -comment $2 /vobs/A/xxx/*.h /vobs/B/xxx/
It works fine. But it will check in all the changes automatically. Is there a way to do the same task but leave the update files in a check out status?
I want to update the file for B from A. Build my programme, and then re-cover the branch. So if the updated files is an check out status, I can do unco later. Well with my command before, everything is checked in. I cann't re-cover my branch then.
Thanks.
As VonC said, it's impossible to prevent "clearfsimport" to do the check in. And he suggested to use a label to recover back.
For me, the branch where I did "clearfsimport" is branched from a label.Let's call it LABEL_01. So I guess I can use that label for recovery. Is there an easy way (one command) to recover the files under /vobs/B/xxx/ to label LABEL_01 ? I want to do it in my bash script, so the less/easy the command is, the better.
Thanks.
After having a look at the man page for clearfsimport, no, it isn't possible to prevent the checkins.
I would set a label before the clearfsimport, and modify the config spec for the new version to be created in a branch (similar to this config spec).
That way, "re-cover" the initial branch would be easy: none of the new version would have been created in it.