loggerhead.history._process_diff should not exist

[Robert Collins]

About 30% of showing a diff in loggerhead with split inventory, is in
loggerhead.history._process_diff, which parses a plain text diff emitted
by bzrlib.

This is more than slightly nutty: we can output the diff in the
appropriate format in the first place rather than post-processing.

-Rob

 affects loggerhead
--
GPG key available at: <http://www.robertcollins.net/keys.txt>.

[Robert Collins (lifeless)]

Turns out most of the time is in bzr grabbing the diffs; all the same, this is a bit fugly layer wise.

[Martin Albisetti (beuno)]

bumping down to low since it's not actually a performance problem

[Michael Hudson-Doyle (mwhudson)]

Oh yes, it's certainly ugly :)

I guess it involves cribbing code from bzrlib/patiencediff/unified_diff.py and rearranging as needed.

[Robert Collins (lifeless)]

On Tue, 2008-12-23 at 19:44 +0000, Martin Albisetti wrote:
> bumping down to low since it's not actually a performance problem
>
> ** Changed in: loggerhead
> Importance: High => Low

30% of diff time isn't a performance issue ?

[Michael Hudson-Doyle (mwhudson)]

Are you taking this comment: "Turns out most of the time is in bzr grabbing the diffs; all the same, this is a bit fugly layer wise." back?

[Robert Collins (lifeless)]

On Wed, 2009-02-25 at 18:46 +0000, Michael Hudson wrote:
> Are you taking this comment: "Turns out most of the time is in bzr
> grabbing the diffs; all the same, this is a bit fugly layer wise." back?

No, just going blind in my old age.

-Rob