hg diff two different files from a certain revision - mercurial

I want to see the difference between filea and fileb as of rev1234, because fileb appears to have been a fork-and-change-and-thereafter-run-two-versions-of-one-library, without a super informative commit message.
I tried hg diff -r 1234 filea fileb, but that just shows me two diffs:
that between filea as of rev1234 and its current version, and
that between fileb as of rev1234 and its current version
I want ONE diff, between filea as of rev1234 and fileb as of rev1234.

I understand that both files are present in r1234?
If r1234 is not your current working dir state, update your working directory to r1234 beforehand: hg update -r1234
Then you do not need any mercurial command but the normal diff command:
diff fileA fileB. When you're done, hg up again to return to normal.

It's roundabout, but you can create a directory, export just those two files from that revision, and then compare them:
mkdir tempdiff
hg archive -r 1234 -I filea -I fileb tempdiff/
diff tempdiff/filea tempdiff/fileb
rm -r tempdiff
A one-command method would be nice, though.

Related

Show if the revision of subrepos changed, or if they are dirty

I have a mercurial repository (main repo) with several sub repositories.
I need a mercurial command to show if the revision of a sub repo changed (including information on old and new revision) in the working copy or if the sub repo state is dirty.
How can I do this?
Most mercurial commands accept the -S or --subrepos flag. Thus by calling hg st -S you get a list of all changed files which include those in the sub-repos, if their state differs from the state recorded in the .hgsubstate file:
$ cd opengfx/
$ hg st
$ hg id
10065545725a tip
$ cd ..
$ hg st -S
M opengfx/.hgtags
M opengfx/Makefile
A opengfx/lang/japanese.lng
$ cat .hgsubstate
785bc42adf236f077333c55c58490cce16367e92 opengfx
As to your wish to obtain the actual revisions, that's AFAIK not possible with one command. However you can always check the status of the individual sub-repos like above or can check them from the main repo by giving mercurial another path to operate on:
$ hg id -R opengfx
10065545725a tip
In order to get the status of each repo compared to what is required by the parent repo, I'd resort to some simple bash:
for i in $(cat .hgsubstate | cut -d\ -f2); do echo $i is at $(hg id -R $i) but parent requires $(cat .hgsubstate | grep $i | cut -d\ -f1); done
which gives output like
opengfx is at 10065545725a tip but parent requires 785bc42adf236f077333c55c58490cce16367e92
In a similar fashion you can also check whether the state is modified by using hg st instead of hg id.

Hg diff with multiple files

I am new to mercurial. So, please excuse my question if it sounds trivial. I am trying to figure out how to do a diff for multiple files. Here is my use case : I made changes to four files. However, I am only interested in seeing the changes I made in two of them (fileA and fileB). I thought something like this would work :
hg diff fileA fileB
But it does not.
You need either the --include(-I) or --exclude(-X) options such as:
hg diff -I fileA -I fileB
hg diff -X *.csv
Remember you can compare specific revisions too
hg diff -r 1234:tip -I fileA
Use hg help diff in the console to see all the options available or look here

Mercurial: Easy way to see changes from last commit

In Mercurial, I can see my current (uncommitted) changes by running
$ hg diff
Fine. But after commit, I sometimes want to see this diff again (i.e., the diff of the last changeset). I know I can achieve this by
$ hg log -l 1
changeset: 1234
tag ...
$ hg diff -c 1234
I'm looking for a way to do this in one line.
Use hg diff -c tip, or hg tip -p (shorter, but works only for tip).
This will work until you pull something, since tip is an alias for the most recent revision to appear in the repo, either by local commit or
pull/push from remote repositories.
You can use relative revision numbers for the --change option:
hg diff -c -1
See https://stackoverflow.com/a/3547662/239247 for more info.
An alternative is to use: hg diff --rev -2:-1
This form has the advantage that it can be used with the status command (e.g. hg st --rev -2:-1), and using it makes it easy to remember what to do when one needs to determine differences between other revision pairs (e.g. hg diff --rev 0:tip).
The answer from Macke is quite helpful, but in my case I didn't want to diff tip.
Thankfully you can also just diff the currently selected comment:
hg diff -c .

Create a new branch, made a lot of changes, how to view list of files changed?

So there was a new branch created where we made some breaking changes to the codebase.
Now we are going to merge, but before that I want to get a list of all the files that were changed in the branch.
How can I get a list of files? I tried:
hg status --change REV
But i'm not sure if that is what I want, since I want all files changed in this branch and not a specific revision in the branch.
BTW, how can I view the revision numbers?
Try with
$ hg status --rev "branch('your-branch')"
to get the changes between the first and the last changeset on the branch (hg status will implicitly use min(branch('your-branch')) and max(branch('your-branch')) when you give it a range of revisions like this).
Since you'll be merging, you should really look at
$ hg status --rev default:your-branch
to see what is changed between the default branch and your-branch. This shows you the modifications done, and filters out any modifications done on the branch due to merges with default.
This is necessary in case your history looks like this:
your-branch: x --- o --- o --- o --- o --- o --- y
/ / /
default: o --- a --- o --- b --- o --- c --- o --- o --- d
where you've already merged default into your branch a couple of times. Merging default into your branch is normal since you want to regularly integrate the latest stuff from that branch to avoid the branches drifting too far away from each other.
But if a new file was introduced on default and later merged up into B, then you don't really want to see that in the hg status output. You will see it if you do
$ hg status --rev a:y
since the file was not present in a, but is present in y. If you do
$ hg status --rev d:y
then you wont see the file in the output, assuming that it's present in both heads.
You write in a comment that you're working Kiln repository. They mean "clone" when they say "branch", but the above can still be adapted for your case. All changesets will be on the default named branch, but that's okay.
Run the following command in your local clone of the "branch" repository:
$ hg bookmark -r tip mybranch
This marks the current tip as the head of mybranch. Then pull all the changesets from the main repository:
$ hg pull https://you.kilnhg.com/Repo/public/Group/Main
You then mark the new tip as the tip of the main repository:
$ hg bookmark -r tip main
You can now run
$ hg status --rev main:mybranch
to see the changes between main and my-branch. If you want to see what you did on the branch itself, the use
$ hg status --rev "::mybranch - ::main"
The ::mybranch part will select changesets that are ancestors of mybranch — this is all your new work, plus old history from before you branched. We remove the old history with - ::main. In older versions of Mercurial, you would use hg log -r -r mybranch:0 -P main.
In cases like this, I prefer to do a test merge from a newly checked-out copy of the repo. This has the advantage that I can see how many conflicts the merge will produce, and I can keep the merge result because I did it in its own copy.
To view the revision numbers, enable the graphlog extension and run:
$ hg log -b your-branch -G
This gives you a nice ASCII graph. This can be handy to quickly look at the graph, but I recommend using TortoiseHg for a cross-platform log viewer:
I had to merge the default branch into my branch to get some fixes, now the commands above shows also files changed because of merges (this files changed after the merge again in the default branch).
Therefore, to get only the correct files I use something like this:
hg log --rev "branch('my-branch') and not merge()" --template '{files}\n' | sed -e 's/ /\n/g' | sort -u
if you have spaces in file names, you can do it this way:
hg log --rev "branch('my-branch') and not merge()" --template '{rev}\0' | xargs -0 -I # hg status -n --change # | sort -u
And to answer your last question, revisions can be shown this way:
hg log --rev "branch('my-branch') and not merge()" --template '{rev}\n'
TIP: I use a hg-alias for this:
[alias]
_lf = ! $HG log --rev "branch(\"$1\") and not merge()" --template '{rev}\0' | xargs -0 -I # hg status -n --change # | sort -u
With mercurial, if you want to get the list of all the files changed in your current branch (changes done of your changeset) you can use these commands: :
hg log --branch $(hg branch) --stat | grep '|' | awk -F\ '{printf ("%s\n", $1)}' | sort -u
Example result:
api/tests/test_my_app.py
docker/run.sh
api/my_app.py
Explanation of the commands:
hg log --branch $(hg branch) --stat
Show revision history of entire repository or files and output diffstat-style summary of changes
hg branch
Show the current branch name
grep '|'
Search for a text pattern, in this case, it is "|"
awk -F\ '{printf ("%s\n", $1)}'
Space separator denotes each field in a record and prints each one in a new line
sort -u
Sort all the printed lines and delete duplicates

How to use `hg cat` from an empty working directory?

I have a repo located at x:/projects/repo1. The working directory has been emptied using hg update null. I want to extract the latest version of some files from there to a local directory.
I tried this:
x:\projects\repo1> hg cat -o c:\sql\%s scripts\*.sql -r tip
I get this error:
scripts\*.sql: No such file in rev 14f07c26178b
The same command works fine if the working directory is not empty. Is there a good reason why this does not work? Or do you know another way of extract some files from there to a local directory?
The hg cat command is for single files. If you want multiple files use the hg archive command, which makes zipfiles or directories full of files. Here's your command:
x:\projects\repo1> hg archive --include scripts\*.sql -r tip c:\sql
It seems that hg cat doesn't support wildcard symbols in paths. So you should use the full file name:
hg cat -r tip scripts/foo.sql
When your working copy is up to date with the tip revision, your shell does wildcard substitution for you.
The hg manifest command also might be helpful for getting tracked file listings.
This answer is to your comment on Andrey's answer:
hg manifest takes a --rev argument that you can use to see the list of all files in your repository:
hg manifest --rev tip
To get the list of files matching a pattern at the tip, use:
hg stat --all --include *.sql --rev tip --no-status
hg stat -A -I *.sql --rev tip -n # using abbreviations.
From there you could redirect the output to a file and edit each line into a hg cat command like in your original question. It appears (to me, at least, having done some experimentation) that hg cat uses the contents of the working directory -- rather than the contents of the repository at the revision specified -- for glob-matching, so once you know the exact name of the file, you can hg cat it at any revision.