So I am giving myself some time to write a git-find script that I can use to feed commits to another tool, like git-graft (or git-cherry-pick).

I don’t really know what I am doing yet, so I want to survey what is available in git-* tools and reuse as much of the available features.

git-find needs to:

  • take as input a commit-range in various formats; I had previously stated that I want something like this:

     <A>                    - all commits from <A> to HEAD
 <A> <B> ...            - individual commits listed
 <A>..<B>               - all commits in given range
 -                      - commit hashes from stdin

… interestingly enough this is almost what can already be done with git-rev-list. git-find should then use git-rev-list to locate the initial set of commits to work on.

  • once we have a commit, we have to decide if that commit matches the selected criteria. That means we have to be able to apply a condition check onto the diffs between a commit and it’s parent. That’s easy, git-diff for looking at lines and files that changed.

  • will it need access to the historical state of the tree?

    There may be some cases where that is useful…

    git-cat-file can take an object in the git store and output the contents, but git-cat-file is very low-level. However, it’s pretty simple to follow the paths to dump a file. cg-admin-cat, from cogito, is a great example on how this can be done.

  • certain types of matches may require that we have to look at the commit message text. Will need to use git-log for that.

I think that’s it. Looks like all the pieces are there. The hardest part, for this bash weenie, will be to represent the conditions in bash.

Why am I not doing this in perl, again?

Anyway, on with the show…

  • first is to find out all the revisions…

     $ git-rev-list HEAD~2..HEAD
 5209eda86363a3ba2e000903ad8de29589b18b58
 24cf6e5847073d50390e0b7950e8e6b5a09103bc
 7b8cf0cf2973cc8df3bdd36b9b36542b1f04d70a
 e14421b9aa85f11853a0dacae09498515daab7b8

  • having those we want to peek at the log…

     $ git-rev-list HEAD~2..HEAD | xargs -n1 git-log --max-count=1 | cat
 <output ommited>

  • now let’s look at the diff for each of those revisions

     git-rev-list HEAD~2..HEAD | while read commit ; do
         git-diff ${commit}~1 $commit
 done

… the rest is really just shell magic.

** … few hours go by … **

So here is the the first implementation git-find.git.

I had to add glob support to git-diff, which took a while to figure out. ;)