Re: What's cooking in git.git (Apr 2014, #01; Fri, 4)

From: Duy Nguyen <pclouds@gmail.com>
Date: Apr 8, 2014, 00:39 UTC
Message-ID: <CACsJy8A0oNRZ45xLvGTDPmDu0ZEAyHbdQBt2gdyo1yUtUeeB_g@mail.gmail.com>
In-Reply-To: <5342EFD4.6020007@kdbg.org>

On Tue, Apr 8, 2014 at 1:35 AM, Johannes Sixt <j6t@kdbg.org> wrote:

Show 15 quoted lines

> Am 05.04.2014 11:19, schrieb Johannes Sixt:
>> Am 04.04.2014 22:58, schrieb Junio C Hamano:
>>> * sz/mingw-index-pack-threaded (2014-03-19) 1 commit
>>>  - Enable index-pack threading in msysgit.
>>>
>>>  What is the status of this topic?  A failure report exists
>>>  ($gmane/245170), and I am aware of Duy's $gmane/245034 but that was
>>>  where the discussion stalled.  Is everybody waiting for everybody
>>>  else to get the discussion unstuck?
>>
>> I still have to cross-check Duy's patch. I'll hopefully get to it in the
>> next days and report back.
>
> The test suite passes with Duy's patch ($gmane/245034), but t5302 fails
> with this patch with a MinGW build.

Is "this patch" the one on 'pu', or mine?

> The patches touch the Cygwin configuration, but I cannot test a Cygwin build.

If no one can test the Cygwin changes, I'm happy to revert it back for safety.

> I have, however, lost track of what the objective of these patches is.
> Is the threaded version significantly faster, and these patches are
> worth it?

It depends on the repo, how much deltas it has, how deep. According to b8a2486 (index-pack: support multithreaded delta resolving - 2012-05-06), going to two cores saves ~20% time. But some actual numbers on Windows would be nice.

-- 
Duy