From: Barret Rhoden <brho@google.com>
Date: Tue, 23 Apr 2019 14:23:44 GMT
Subject: Re: [PATCH v6 0/6] blame: add the ability to ignore commits
Message-ID: <cc1466bc-0610-784b-e57b-8612c2e8569f@google.com>
In-Reply-To: <20190422222647.48628-1-michael@platin.gs>

On 4/22/19 6:26 PM, michael@platin.gs wrote:
> From: Michael Platings <michael@platin.gs>
> 
> Hi Barret,
> 
> This patch is on top of your patch v6 4/6.

Thanks, I'll take a look.  I was working on taking your old version and 
integrating it with my v6 6/6.  That way it gets the 
origin-fingerprint-filling code and can be easily compared to my 6/6 style.

[snip]

> 
> To address reordered lines I suggest a combination of this algorithm and your
> algorithm - in the first path my algorithm tries to match lines within a
> single chunk, and in the second pass your algorithm tries to find matches for
> unblamed lines out of order and outside their chunk.

I was thinking something similar.  Yesterday I did this with your older 
patch set - applied on my 6/6.  Two passes, one with your fuzzy matcher, 
then if we didn't find anything, do a scan of the entire parent (as my 
6/6 does now).

This approached worked for the cases I had (e.g. "header reordering"). 
I ran into an issue last night where your scan was finding matches where 
it shouldn't - might have been an issue with how I hooked it up.  I'll 
try your latest code and see how it goes.

Thanks,

Barret