Re: [RFC] Add bad-branch-first option for git-bisect

[Christian Couder <christian.couder@gmail.com>]

On Mon, Jan 24, 2011 at 11:30 AM, Shuang He <shuang.he@intel.com> wrote:
> On 2011/1/24 17:53, Christian Couder wrote:
>>
>> Hi,
>>
>> On Mon, Jan 24, 2011 at 3:03 AM, Shuang He<shuang.he@intel.com>  wrote:
>>>
>>> Hi
>>>     The default git-bisect algorithm will jump around the commit tree,
>>> on the purpose of taking least steps to find the first culprit commit.
>>> We may find it sometime would locate a old culprit commit that we're not
>>> concerned about anymore.
>>
>> Yes, it can be a problem.
>
> I'm honored to be given so much comment :)
> Thank you

I am honored by your interest in git bisect and the fact that you
provided a patch :-)
Thanks!


>> If the quality of these branches is too bad, I think they should not
>> have been merged in the first place.
>> If they are not merged (and not marked as good), then git bisect will
>> not look at them, since it will look only at commits that are
>> ancestors of the bad commit it is given.
>>
>> Or if one is merged but it causes too many problems, then perhaps a
>> replacement commit could be used to unmerge the branch.
>>
>> Another possibility is to have in a file a list of commits that are
>> the last commits on these branches before the merge commits, and do a:
>>
>> git bisect good $(cat good_commits_file.txt)
>>
>> at the beginning of each bisection.
>>
>> So I think the long term solution in this case is not what your are
>> suggesting.
>
> Yeah, I agree that the issue I addressed above will not be a problem if all
> those branches are maintained very well.
> Actually we've implemented a automated bisect system for Intel Linux
> Graphics Driver Project, and so we'd like the system
> helps us to locate issue in an more automatic way when branches are not
> maintained as good as expected.

I think there is always a price to pay when you bisect if the branches
are not well maintained.
Maybe your algorithm could help in some cases, but my opinion is that
there will probably still be many problems and a human will often have
to take a look.

>>>         2. Some of those branches may not synchronized with main
>>> branch in time.  Say feature1 is broken when feature2 branch is created,
>>> and
>>> feature1 is fixed just a moment later after feature2 branch is created,
>>> and when feature2's development is done, and developer want to merge
>>> feature2 branch back to master branch, feature2 will be firstly
>>> synchronized to master branch tip, then merge into master.  For the same
>>> reason addressed in issue 1, this will also lead git-bisect into wrong
>>> direction.
>>
>> I am not sure what you mean by " feature2 will be firstly synchronized
>> to master branch tip", and I think this should mean a rebase that
>> would fix the bug if feature1 has already been merged into the master
>> branch.
>>
>> But anyway in this case, I think that git bisect will find that the
>> first bad commit is the last commit in the branch, just before it was
>> merged. And by looking at the branch graph it should be quite easy to
>> understand what happened.

Now I think I was wrong here, as git bisect will probably find that
the first commit in the branch (not the last one) is the first bad
commit.

[...]

>> - the name "bisectbadbranchfirst" seems wrong to me, because git
>> branches are just some special tags; "firstparentsonly" would be a
>> better name,
>
> It's recursively applying bad branch first algorithm, not just constantly
> stick to first parent.
> Given this condition:
>    A -> B -> C -> D -> E -> F -> G -> H   (master)
>         \ a  -> b -> c -> d -> e /  (feature 1)
>              \ x -> y -> z/      (feature 2)
> start with H as bad commit, and A as good commit, if y is the target bad
> commit. bad-branch-first algorithm will do it like this:
>    1. In first round stick to master branch, so it will locate G as first
> bad commit
>    2. In second round stick to feature1 branch, then it will locate d as
> first bad commit
>    3. In third round stick to feature2 branch, then it will finally locate y
> as first bad commit
> So you could see, it's always sticking to branch where current bad commit
> sit

I see. It is interesting, but why not develop a "firstparentsonly"
algorithm first?

As Avery explains in his email, it is already interesting to have a
"firstparentsonly" algorithm because some people are only interested
to know from which branch the bug comes from.
When they know that, they can just contact the relevant people and be
done with it.

And when we have a "firstparentsonly" algorithm, then your algorithm
could be just a script that repeatedly uses git bisect with the
"firstparentsonly" algorithm. And this script might be integrated in
the "contrib" directory if it not considered important to be
integrated as an algorithm into git bisect.

Thanks,
Christian.