Re: merge-recur status

From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Aug 10, 2006, 07:54 UTC
Message-ID: <Pine.LNX.4.63.0608100949250.13885@wbgn013.biozentrum.uni-wuerzburg.de>
In-Reply-To: <20060810062914.GA5192@c165.ib.student.liu.se>

Hi,

On Thu, 10 Aug 2006, Fredrik Kuivinen wrote:

> I just want to say that you have made a great work with porting
> merge-recursive to C!

Thank you! I was feeling bad a little, because you were the creator of it, 
and I just bastardized it. But given my plans with MinGW, I needed to get 
rid of Python (it exists on Python, but has all kinds of path conversion 
problems).

Show 10 quoted lines

> I ran merge-recur through a couple of test merges that I used to test
> merge-recursive with. It is mostly merge cases people have posted to
> the mailing list, but also some home made ones. For some of them I get
> segmentation faults, see the log below. The first three come from
> Linus kernel tree. The last one,
> 44583d380d189095fa959ec8ba87f0cb6deb15f5, is from Thomas Gleixner's
> historical Linux kernel repository. I haven't seen "fatal: fatal:
> cache changed flush_cache();" before...
> 
> Let me know if you can't reproduce some them.

I could reproduce already yesterday ;-) The patch "merge-recur: do not die unnecessarily" I sent out earlier gets rid of this error.

But yes, I want to test -recur with the linux kernel on the weekend.

Ciao, Dscho

P.S.: Pavel, thanks for the "A dedicated programmer with good C and Python skills could rewrite git-merge-recursive.py in C in 2 days, I believe. Add a few days of bug fixing, of course." underestimation. I do not know if I had started fiddling with it if I had known how involved it is.