This is basically the same question as Temporarily switch working copy to a specific Git commit, but for Mercurial.
Say revision 500 is the last revision I've committed in Mercurial. I've worked some more, and realized something went wrong in an earlier revision. So, I've "git stashed", that is, hg shelved my current changes, so I am at a clean revision 500.
Now what I want to do is checkout revision 499, rebuild software, see if it still has error; if it does, checkout revision 498, rebuild software, see if it still has error; if it does, revision 497, rebuild software, see if it still has error; etc etc until I figure out which revision introduced the error.
Once I'm done with that, I want to go back to revision 500, unshelve my previous work, add to that a fix, and then make a new commit (which would be revision 501). That means when I go back in history I don't want to reset anything, nor stage anything from there.
In git I can just do git checkout HASH to go back in history, and once done, git checkout master to go back to the tip. Which commands should I use in Mercurial?
You can go to older revisions using hg update:
hg update 499 # The short id number or the full hash
or alternatively if you want to step to the parent changeset without needing to know the id, you can use a mercurial revset:
hg update .~1
once you have identified your issue you can simply:
hg update tip
and unshelve your changes, add your fix and commit.
I recommend you read the Mercurial: The Definitive Guide. Your question very basic / entry level mercurial usage.
Edit: the link to Mercurial: Definitive Guide is:
http://hgbook.red-bean.com/read/
or the 1.9 version of the boook:
https://book.mercurial-scm.org/read/index.html
Related
Is there a way to edit a commit message in Mercurial on a commit after other commits have occured using TortoiseHg? I've read these posts:
How to edit incorrect commit message in Mercurial?
Mercurial: how to amend the last commit?
and have been able to update a "regular" commit message when it is the latest commit on a branch (using TortoiseHg). However, I haven't been able to figure out how to edit a commit message when other commits have occurred after the one I want to edit. It always edits the last commit.
Based on Ed Cottrell's comment, I did a test where I made two commits without pushing to the central repo, and I still have the same issue - only the last commit message can be edited.
EDIT: I should clarify that I am looking to update a changeset that has been pushed.
Histedit extension (bundled with TortoiseHG now) has a mess command for changing the commit message of historical changesets.
Unfortunately, this command is not supported by the TortoiseHG GUI so you need to run the command from command line.
As long as the change in question is local and hasn't been pushed anywhere, it is possible.
The commit message is used to compute the globally unique hash id that is used for all repositories to determine whether or not they already have a changeset. If you change the commit message, you change the unique hash id and every repo will see it as a "new" changeset. All other repositories that had the old changeset will try to get the new one and ask that you merge it with itself.... This is not a good thing, so the short answer to your question is "don't do it".
If you could definitively purge that change from all other repos, so that only the local copy is left you could essentially get to the "draft" state. Note that if any repo has the "old" changeset, it will be pushed to the central repo someday and cause the mess that we are trying to avoid.
If the changeset is still local (e.g. in draft status), you can use hg commit --amend if it is the parent of the working directory.
If there are changes after it, I would use mq and hg qimport all the changes down to and including the one where you want to edit the commit message. hg qpop -a and then hg qpush to get to the patch that represents the changeset you want to edit. Then hg qrefresh -e and make your changes. Then just hg qfin -a and you should be good to go.
The advice from Edward is good — if you've pushed your changes to another repository, you should consider them set in stone and not update the commit message or any other aspect of them.
However, we're working on changing this in Mercurial. There is an experimental extension that will allow you to do more extensive history editing and push those edits to other repositories. It is called the Evolve Extension and it enables some behavior that is partly in the core of Mercurial and partly outside core.
When using evolve, you can edit the second-to-last commit message like this
$ hg update .^
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg commit --amend -m 'new commit message'
1 new unstable changesets
$ hg stabilize
more:[5] old tip changeset
atop:[6] new commit message
The extension allows you to do this as long as the changesets are in the draft phase. To keep them in the draft phase after pushing them somewhere, the repository you push to need to be configured as a non-publishing repository. You can read more about this in the Changeset Evolution Documentation.
Many times it happens that I have few commits on my local Hg repository which I don't want to push and sometimes I want to remove the local branch altogether. But I cannot rollback more than one commit which leaves me no choice than creating a new clone and download the whole repository again. This feels stupid, since if I could just delete my local branch which has not affected the remote repository in anyway, then I wouldn't have to create and setup a new clone. So, is it how it is in Mercurial or is there some way to discard a local branch?
Thanks!
If you enable the mq extension (bundled with Mercurial), you can use hg strip. Be careful, though, as this will modify the history of your repository. The safe method is to clone your repository up to the revision preceding the creation of the branch you want to discard, then to pull the remaining changesets that you want to keep.
I know its too late but it may be useful for any one:
If your branch is not pushed yet.
First rollback changes hg rollback only if you have done commit but
not yet pushed
Second run hg update --clean
Third run hg branch any-existing-branch
Fourth run hg pull -u
If you find yourself doing this often perhaps you should be using bookmarks instead of named branches. http://stevelosh.com/blog/2009/08/a-guide-to-branching-in-mercurial/
When getting the latest code from a Mercurial repo on the command line, if there are changesets that need to be merged Mercurial raises a warning:
hg up
abort: crosses branches (merge branches or use --check to force update)
This is what I expect, and from the Mercurial book it says "Mercurial is telling us that the hg update command won't do a merge; it won't update the working directory when it thinks we might want to do a merge, unless we force it to do so." At this point I know I need to merge.
How can I get the same behaviour using TortoiseHg? When I hit "Update", it happily updates me to the most recent changeset. Is there a way to warn me that a merge is probably needed? The "Always merge (when possible)" option seems to only apply when you have uncommitted changes.
The reason you get an error from hg update on the command-line is that it doesn't know which revision to pick. There are 2 divergent default heads.
If you were to execute hg update -r <specific rev>, the command completes without error.
When using TortoiseHg, you update by:
right-clicking a specific changeset
selecting Update...
This translates to hg update -r <rev>, so there is no error.
Using TortoiseHg, you always have the revision graph in front of you. The graph shows when newly pulled changesets create a new head.
The Mercurial abort you are seeing is occurring because you have outstanding (ie: non-committed) changes in your working directory, are trying to perform an update, and Mercurial has decided that you should probably perform a merge instead.
TortoiseHg should also warn you about this, but it will do so in a different way. It will spawn a dialogue that asks if you want to Discard, Merge, or Shelve your outstanding changes. This is what it looks like in TortoiseHg v2.X.X, but it should be similar in v1.1.X:
If you're not seeing this in TortoiseHg, you may not have any outstanding changes. Try it again - are you seeing these options?
I want to completely delete a Mercurial commit as if it was never entered in the repository and move back to my prior commit.
Is this possible?
If it was your last commit and you haven't pushed it anywhere, you can do that with rollback. Otherwise, no. Not really. Time to change your passwords.
Edit: It has been pointed out that you can clone from an older revision and merge in the changes you want to keep. That's also true, unless you have pushed it to a repo you don't control. Once you push, your data is very likely to be very hard to get back.
You can try to remove mq info about your commit.
For this you need to go File->Settings->Extensions.
There check mq and restart gui.
After that just right click on unneeded commit and
ModifyHistory->Strip
To edit the history I would use the Histedit Extension extension.
hg histedit 45:c3a3a271d11c
However keep in mind this only makes sense in a situation where you have not yet pushed the commits to the public repository, you own the public repository and/or you can account for all the clones out there. If you receive the following error:
abort: can't rebase immutable changeset 43ab8134e7af
It means that Mecurial thinks this is a public changeset (see phases) that has already been pushed - you can force it to be a draft again doing:
hg phase -f -d 45:c3a3a271d11c
I encounter this fairly often. I make a commit and then pull to push. But then there is something incoming that makes my newly made commit unnecessary. A plain hg rollback isn't enough because it only undoes the pull...
This is the thing to do:
hg strip <rev>
Things are painless when you don't push your changesets anywhere.
If it's more than one commit and/or you already pushed it somewhere else, you can clone your repository and specify the last changeset that should be cloned.
See my answer here how to do this:
Mercurial: Fix a borked history
If you only committed locally and didn't push, you can just create a clone locally (as described in my link) and you're done.
If you already pushed to some remote repository, you would have to replace that with your clone.
Of course it depends if you are able (or allowed) to do this.
You can use "hg backout" to do a reverse merge basically. All options are discussed in the freely available book "Mercurial: The Definitive Guide":
http://hgbook.red-bean.com/read/finding-and-fixing-mistakes.html
If using tortoise you can use modify history > strip...
Yes. Unless I am mistaken, as of v2.3 (rel. 2012/08/01) you can use the HisteditExtension with a drop command to drop a commit, along with strip or backout to remove changes.
A simple Google search on the feature: https://www.google.com/webhp#q=histedit+drop
In 2022 I do use evolve extension. It is one of the best extensions for this purpose.
To prune unwanted changeset, if you for example did a quick hack to get the code working:
$ echo 'debug hack' >> file1.c
$ hg commit -m 'debug hack'
Now you have a proper patch you can do hg prune .:
$ hg prune .
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
working directory is now at 2a39221aaebb
1 changesets pruned
If you push the change to the remote repository you will find only obsolescence markers:
$ hg push
searching for changes
no changes found
remote: 1 new obsolescence markers
To check the changes to your local repo you can pull from the remote one:
$ hg pull
pulling from ssh://userid#server/repo
searching for changes
no changes found
I comitted few times with mercurial and pushed my commits. Nobody else pulled them. How can I "throw away" these commits to restore my local and remote repo (both repo and working dir) to a specified commit?
Thank you.
The quickest way is to just clone your local repo to the last 'good' revision, and discard the old one.
This only works if those revisions are the last n in the history.
You can copy the [paths] section of the old repo's .hg/hgrc file to the new one if you want to continue pushing to a third 'master' repo.
You can use $hg revert to revert your code to a previous revision.
If you type $hg help revert into your command line you'll get a list of ways you can revert.
These options include:
--all
--date
--rev
--rev will likely be the one you want since it reverts back to a specific revision.
If the revision you want to revert to is the same as another repo or the same as the current tip, the easiest way then could be to just re-clone that repo.