Is there a canonical explanation why the same parent revision is listed twice for a given mercurial changeset resulting from a merge.
Unfortunately, we do not have a chance to tell, what exactly caused this situation in the observed case. It might be that the situation was caused by trying to perform an empty merge operation.
Any hints appreciated!
We found out a possible way to cause this kind of situation...
The situation can easily be reproduced with TortoiseHG by selecting your local version and "Merge with local" checking the option "Discard all changes from merge target (other) revision".
This does not result in calling hg merge (as one could assume), but hg debugsetparents. So we shot ourselves in the foot by forcing Mercurial to set both parents to the same revision.
Related
I made some changes to a file and committed it. (In fact there were several commits).
Then I wanted to revert to the earlier version and lose all those changes.
I did something like:
hg update -r nnn where nnn was the reversion number of the changeset I wanted to go back to.
That worked. I was happy.
Then, later, I had to push my local repository to the remote. But when I did hg push I got a message about there being two heads on this branch and one of them not being known to the remote repositiory. It suggested I merge before pushing. (I think).
I googled this and found a page that suggested I do "hg merge". I did that. Now the resultant file is back to where I started. I.e. it contains all the changes I wanted to throw away.
Where did i go wrong?
EDIT:
I have found this post Mercurial — revert back to old version and continue from there
where it says:
If later you commit, you will effectively create a new branch. Then
you might continue working only on this branch or eventually merge the
existing one into it.
That sounds like my case. Something went wrong at the merging stage it seems. Was I on the wrong branch when I did "hg merge"?
You're past this point now but if it happens again, and it's just a single file you want to revert then consider:
hg revert --rev REVISION_YOU_LIKED path/to/just/one/file.txt
That doesn't update you whole repository to a different revision, and it doesn't create any commits. It just takes a single file in your working directory and makes it look like it used to. After doing that you can just commit and you're set.
That's not the way to go if you want to undo all the changes you've made to all files, but for reverting a single file use revert and avoid multiple heads and merging entirely.
No, nothing went wrong at the merge stage – Mercurial did exactly what you asked it to...
What merge means is that you take the changes on your current branch, and the changes on the 'other' branch, and you merge them. Since your original changes were in the 'other' branch, Mercurial carefully merged them back into your current branch.
What you needed to do was to discard the 'other' branch. There are various ways of doing that. The Mercurial help pages discuss the various techniques, but there are pointers in other SO questions: see for example Discard a local branch in Mercurial before it is pushed and Remove experimental branch.
(Edit) Afterthought: the reason you got a warning about there being two heads on the branch is because having two heads is often a temporary situation, so pushing them to a remote repository is something you don't want to do accidentally. Resolutions are (i) you did mean to push them, so use --force to create two heads in the remote repository; (ii) ooops!, you meant to merge them before pushing, so do that; or (iii) ooops!, you'd abandoned the 'other' one, so get rid of it. Your case was (iii).
As I understand it, you can't really fix a comment in Hg. So what I would like to do instead is re-push the exact same changes (or at least "touch" the same files and commit & push again).
The reason this is necessary is because we have a bug tracking and build system that relies on specific comment patterns, and we need to make sure the right files get included in the build, but if I forget to update the bug # in my comment from my last commit, and I accidentally commit and push it under the wrong # because i'm overzealous, how can I re-push those same files again without manually going into each one and adding a space or line break just to create a diff?
To clarify, I can't "rollback" or something; it's already been pushed with the wrong message.
As far as I know, current Mercurial features provide no support for this. After the changeset has been pushed, there's little you can do to un-push it, besides stripping it from the server repo and any other developer's repo.
I guess you you should ask those who set up this workflow in your shop; they should've come up with some exception handlers for it.
We usually just ignore issues like this, and close the bug by hand, making sure the bug links to the correct changeset. If the changeset is really messed up (usually this means bad changes, not a malformed commit message), we resort to stripping.
Since your change has already been pushed you can't use a simple fix, like "hg commit --amend", but you can do something similar. Basically, the following commands re-do the commit with Mercurial's help:
CSET=...the changeset to re-do...
hg up -r "p1($CSET)" # Update the working directory to the parent revision
hg log -r "$CSET" -p > changes.patch
hg import --no-commit changes.patch
hg commit # And use the appropriate commit message.
Then, merge and push.
The only way that I could think of doing this is to commit two more changes, one would be an hg backout of the incorrect revision and the other would be an hg backout of that revision with the corrected comment.
I don't like that idea though and wouldn't recommend it if there was any way to fix the problem in your bug tracking system.
This question already has answers here:
Mercurial: Merging one file between branches in one repo
(5 answers)
Closed 2 years ago.
Reading up on Mercurial, it seems to always branch and merge the complete repositories.
Is it possible to just merge some files from one branch to another? (For example I may only wish to merge in the files that fix a given bug.)
Likewise can I cherry pick some change sets, but still have a correct merge record, so if a complete merge is done later it is correct?
I am coming from a perforce “mindset” so may be thinking about this the wrong way.
Yes, Mercurial always branches and merges the whole tree. You don't have the "flexibility" that something like perforce gives you to select individual files for a merge. This is a good thing (trust me). Changesets are atomic (you can't split them) and immutable (you can't change them). Hence this needs a little bit of a mindset change.
Changesets should be targetted at one task, and one task only. If you're fixing a bug, nothing else goes in the changeset apart from the bug fix. You've then got a changeset which documents that bug fix, and you haven't got the problem of wanting to split it. It wouldn't make sense to want to. Half a bug fix is often worse than no bug fix.
When it comes to merging that there's a couple of options:
One school of thought says you should go back to where the bug was introduced. Fix it. Commit (making a small anonymous branch), and merge that forward onto whatever head you want it on (dev, stable, release, whatever). This isn't always practical though.
Another method is fixing the bug in the release branch, and then merging to the development branch. This normally works well.
Alternatively you could fix it at the head of your development branch, but then if you merge it onto your release branch you'll bring over all your development changes. This is where graft (new in 2.0) and the older transplant extension come into play. They allow you to "cherry-pick" a single or range of changesets from another branch and place them on another branch.
Reading up on Mercurial, it seems to always branch and merge the
complete repositories.
Yes
Is it possible to just merge some files from one branch to another? (For example I may only wish to merge in the files that fix a given bug.)
Just touch only "some files" in needed changeset and merge branch with this changeset in head with another branch or transplant in any time
Likewise can I cherry pick some change sets, but still have a correct merge record, so if I complete merge is done later it is correct?
Yes, you can transplant| any changesets to another branch, applied state will be remembered and changes will not be duplicated on final merge
Changes were made to our .vcproj to fix an issue on the build machine (changeset 1700). Later, a developer merged his changes (changes 1710 through 1715) into the trunk, but the mercurial auto-merge overwrote the changes from 1700. I assume this happened because he chose the wrong branch as the "parent" of the merge (see part 2 of the question).
1) What is the "correct" mercurial way to fix this issue, considering out of all the merged files, only one file was merged incorrectly, and
2) what should the developer have done differently in order to make sure this didn't occur? Are there ways we can enforce the "correct" way?
Edit: I probably wasn't clear enough on what happened. Developer A modified a line in our .vcproj file that removed an option for the compiler. His check-in became changeset 1700. Developer B, working from a previous parent (let's say changeset 1690), made some changes to completely different parts of the project, but he did touch the .vcproj file (just not anywhere near the changes made by Developer A). When Developer B merged his changes (becoming changes 1710 through 1715), the merge process overwrote the changes from 1700.
To fix this, I just re-modified the .vcproj file to include the change again, and checked it in. I just wanted to know why Mercurial thought that it shouldn't keep the changes in 1700, and whether or not there was an "official" way to fix this.
Edit the second: Developer B swears up and down that Mercurial merged the .vcproj file without prompting him for conflict resolution, but it is of course possible that he's just misremembering, in which case this whole exercise is academic.
I will address the 2nd part of you question first...
If there is a conflict, the automated merge tools should force the programmer to decide how the merge happens. But the general assumption is that a conflict will involve two edits to the same set of lines. If somehow a conflict arises because of edits to lines that are not close to each other the automated merge will blithely choose both of the edits and a bug will appear.
The general case of a merge tool always merging properly is very hard to solve, and really can't be with current technology. Here is an example of what I mean from C:
int i; // Someone replaces this with 'short i' in one changeset stating
// that a short is more efficient.
// ... lots of code;
// Someone else replaces all the 65000s with 100000s in another changeset,
// saying that more precision is needed.
for (i = 0; i < 65000; ++i) {
integral_approximation_piece(start + i/65000.0, end + (i + 1) / 65000.0);
}
No merge tool is going to catch this kind of conflict. The tool would have to actually compile the code to see that those two parts of the code have anything to do with eachother, and while that would likely be enough in this case, I can construct an example that would require the code to be run and the results examined to catch the conflict.
This means that what you really ought to do is rigorously test your code after a merge, just like you should after any other change. The vast majority of merges will result in obvious conflicts that a developer will have to resolve (even though that resolution is often fairly obvious), or will merge cleanly. But the very few merges that don't fit either category can't easily be handled in an automated fashion.
This can also be fixed by development practices that encourage locality. For example a coding standard that states "Variables should be declared near where they're used.".
I'm guessing that .vcproj files are particularly prone to this problem since they are not well understood by developers and so if conflicts do appear they will not be sure what to do with them. My guess is that this happened and your developer simply did a revert back to the revision (s)he checked in.
As for part 1...
What to do in this case depends a lot on your development process. You can either strip the merge changeset out and redo it, though that won't work very well if lots of people have already pulled it, and it will work especially poorly if there are lots of changesets that have already been checked in that are based on the merge changeset.
You can also check in a new change that fixes the problem with the merge.
Those are basically your two options.
The tone of your post seems to me to indicate that you may have some politics surrounding this issue in your organization, and people are blaming this error on the frequent merges of Mercurial. So I will point out that any change control system can have this problem. In the case of Subversion, for example, every time a developer does an update while they have outstanding changes in their working directory, they are doing a merge, and this kind of problem can arise with any merge.
In mercurial a merge doesn't have a single parent, it by definition has two and only two parents. When someone is merging they're making two choices:
What two changesets will constitute the two changes
Which of those changesets will be the left-parent and which will be the right-parent
Of those two questions the first is very important, and the second barely matters at all, though it took me a while to come to understand that.
You select the left-parent by using hg update X. That changes the output of hg parents (or in newer versions hg summary) and essentially determines what's in your working directory before the merge.
You select the right-parent by using hg merge Y. That says merge X (the working directory's parent) with changeset Y. As a special case, if there are only two heads in your repository and your parent is already one of them then Y will default to the the other.
I'd have to see your resulting graph to know just what the developer did, but it's possible he didn't update to one head or another before invoking merge, which would have him merging one head with some point back in history.
If your developer picked the right parents for the merge then the left vs. right doesn't much matter -- the only real difference is that when one uses hg diff or hg log -p or some other command that shows the patch for a merge changeset, it's displayed relative to the left-parent. That's, however, mostly a factor in display only. Functionally they're pretty much identical.
Assuming your developer picked the right changesets then what he should have done was test the result of the merge before committing it. Merging is software development, not an annoying VCS side effect, and not testing before committing is the error.
Fixing
To fix this, just re-do the merge correctly. Use hg update to set one parent, use hg merge to pick the other. Make sure your current working directory is correct and then commit. You can get rid of his bad merge using something like hg strip or better, just close down his branch with hg commit --close-branch after updating to it.
Avoiding
You say "mercurial auto-merge", but mercurial doesn't really auto-merge. It does a premerge which is an extremely cautious combination of obvious changes, but it's so careful it won't even merge for you if each merge parent adds code in the same region because it can't know which block of code you'd rather have first.
You can disable this premerge entirely or on a file-by-file basis using the merge tool configuration options:
https://www.mercurial-scm.org/wiki/MergeToolConfiguration?highlight=premerge
My company is switching to Mercurial, and we're coming from Subversion.
We're noticing that we're having to do a LOT of merging in our workflow. For instance, if I change a file, commit, pull, update, push, and then my co-worker changes a file, commits, pulls, and updates, he gets a "crosses branches" error and has to do an hg merge. We're having to do this pretty much every single time we want to push to our central repository.
Is something wrong with our workflow?? It seems wrong that in our history for a given file there are going to be a ton of history entries that say "Merging with [changeset id]" "Merging with [changset id]."
Is this just the way it is? Or are we doing something wrong?
There's nothing wrong with this. The vast majority of merges should be automatic. You did create two heads when you both made changes stemming from the same revision and going in divergent directions - your changes might or might not conflict.
If you want to eliminate the "merge" changesets (which aren't actually a problem), you can change/pull/rebase/commit/push instead of change/commit/pull/merge/commit. In other words, before committing your changes, rebase them to the new tip.
If the merges are being executed without manual merge resolution, then I would say mercurial and your workflow are behaving as designed.
It is common as you really do need to get your repos in a consistent state. One thing that speeds it up is instead of
hg pull; hg update
use fetch
hg fetch
This will intelligently do a pull and than either an update or a merge. It comes with mercurial so it's basically a matter of editing your .hgrc to add a line like so:
[extensions]
hgext.fetch=
If the merge goes cleanly you won't even notice it happening. That was a big help in my workflow.