threads / discuss / 21050

is smarter bisect possible?

Subject: is smarter bisect possible?

## tl;dr

2 messages between Sep 25, 2009 and Sep 25, 2009.

replies: 1people: 2as markdown or json

Arkadiusz Miśkiewicz· Sep 25, 2009, 09:37 UTC · lore

Hi,

I have such problem. Linus kernel tree merged some feature (94e0fb086fc5663c38bbc0fe86d698be8314f82f) that was developed in .31rc9 era.

Current Linus kernel tree with that feature merged shows some regression. Tree just before that merge doesn't show regression.

Now if I start to bisect just that feature (good b7f21bb2e23b4fec16b448a34889f467465be659, bad 94e0fb086fc5663c38bbc0fe86d698be8314f82f) then git gets me whole tree back to .31rc9 era which... doesn't show any problems.

This means that only combined current Linus tree and that feature shows a problem.

Is it possible to make git bisect test current tree + parts of that merge instead going back to .31rc9 era with the whole tree?

-- 
Arkadiusz Miśkiewicz        PLD/Linux Team
arekm / maven.pl            http://ftp.pld-linux.org/

Johannes Sixt· Sep 25, 2009, 07:59 UTC · re: Arkadiusz Miśkiewicz · lore

Re: is smarter bisect possible?

Arkadiusz Miśkiewicz schrieb:

Show 13 quoted lines

> I have such problem. Linus kernel tree merged some feature 
> (94e0fb086fc5663c38bbc0fe86d698be8314f82f) that was developed in .31rc9 era.
> 
> Current Linus kernel tree with that feature merged shows some regression. 
> Tree just before that merge doesn't show regression.
> 
> Now if I start to bisect just that feature (good 
> b7f21bb2e23b4fec16b448a34889f467465be659, bad 
> 94e0fb086fc5663c38bbc0fe86d698be8314f82f) then git gets me whole tree back 
> to .31rc9 era which... doesn't show any problems.
> 
> This means that only combined current Linus tree and that feature shows a 
> problem.

Try this:

http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#bisect-merges

-- Hannes