Re: Have git SHA1 ID's changed in the last 5 years?

[David Kastrup]

Benkő Pál <address@hidden> writes:

> 2014-07-27 9:54 GMT+02:00 David Kastrup <address@hidden>:
>> Benkő Pál <address@hidden> writes:
>>
>>>> Okay, so I'm not crazy.  I happened to keep an old repo from
>>>> four years ago, and as you can see from the attached image,
>>>> the two histories are not the same, and neither are the
>>>> commit SHA1 IDs...  How did this happen, and could it happen
>>>> again?
>>>
>>> could you track down the first divergence and/or give more info
>>> about those repositories?  I've tried to check the commits
>>> shown in your attached picture, but couldn't find any.
>>> in fact the authors are quite suspicious, in my repo there's
>>> no commit from fred, neither any from 1999-10-05.
>>
>> In ~/.mailmap I see a line
>>
>> Han-Wen Nienhuys and Jan Nieuwenhuizen <fred> # see note below
>>
>> but in my repository no commits authored by fred, so this suggests to me
>> someone rewriting the history of the repository at some point of time in
>> order to get rid of nonsensical author attributions.
>
> well, the difference in dates is still to be explained.  1.3.0 happened on
> 1999-10-26 with 3a0e9efb7f067e5b334ba0596b95e15d96d7cc49,
> matching neither of Mark's repos.

Well, the solution would appear to be in the archives indeed.  Check out
<URL:http://lists.gnu.org/archive/html/lilypond-devel/2010-03/msg00303.html>
and followups.

So it would appear that indeed the old SHA1 sums are in need of repair.
If anybody does that, I'd strongly recommend quoting several more
characters of the full SHA1 since obviously the LilyPond repository is
growing all the time and thus short identifiers that were at some point
of time unambiguous can stop being so.

-- 
David Kastrup