From: Derrick Stolee <stolee@gmail.com>
Date: Wed, 18 Sep 2019 12:26:41 GMT
Subject: Re: [RFC/PATCH] commit-graph: generation v5 (backward compatible date ceiling)
Message-ID: <8fba9eb7-934f-9b4e-f898-bab099b95ede@gmail.com>
In-Reply-To: <86o8ziatb2.fsf_-_@gmail.com>

On 9/18/2019 4:43 AM, Jakub Narebski wrote:
> Derrick Stolee <stolee@gmail.com> writes:
>> On 6/25/2019 3:51 AM, Jakub Narebski wrote:
>>> Jakub Narebski <jnareb@gmail.com> writes:
>>>> Derrick Stolee <stolee@gmail.com> writes:
> [...]
>>>> O.K., so the "generation number v2 (legacy)" would be incremental and
>>>> backward-compatibile in use (though not in generation and validation).
> [...]
>>>> Do you have benchmark for this "monotonically offset corrected commit
>>>> date" generation number in https://github.com/derrickstolee/git/commits/reach-perf
>>>> and https://github.com/derrickstolee/gen-test ?
>>>
>>> I guess this will have to wait...
>>
>> I have not had time to revisit this topic and re-run performance
>> numbers, sorry.
> 
> I have created pull requests against `reach-perf` branch of
> derrickstolee/git repository[1], and companion pull request against
> gen-test repository[2] with proposed prototype of backward-compatible
> corrected commit date (with monotonic offsets).
> 
> Could you please run the tests for this generation number v5?

I'll try to get to this, but it may take a few days.

> I was not
> able to do so.  It was my first time trying to compile Git on MS
> Windows, and while there were no problems compiling `master` (well,
> except for compilation taking a long time), I was unable to do it for
> `reach-perf` branch because of independent of change compilation errors.

I don't think I tested this on Windows. I do most of my performance tests
in Linux, especially when presenting numbers to the mailing list. The
gen-test repo has a bunch of shell scripts that I used for testing in
that environment.

Thanks,
-Stolee