Re: Re: [COMMITTERS] pgsql: Remove outdated comments from the regression test files.

On Mon, Nov 29, 2010 at 7:45 AM, Magnus Hagander <magnus(at)hagander(dot)net> wrote:
>> Let's do both: "This fixes the bug introduced by the foobar patch from Sep
>> 12th (git commitid a2c23897bc).
>>
>> I like to see the date of the referred patch in the commit message, to get
>> an immediate idea of whether it was a 5-year old change or something from
>> the previous day. But the commitid is also nice so you can immediately
>> copy-paste that without reading through the old commit logs.
>
> +1.
>
> Having the git id is very useful, and putting the date in makes it no
> *less* informational than what we had before, if/when we move away
> from git and it's hashes.

That works for me. But should we make a practice of writing the
ENTIRE SHA-ID rather than an abbreviated form, so that we could more
easily replace 'em later if need be? I think that would be a good
idea for other reasons anyway - it's always possible - though
admittedly unlikely - that a later commit could introduce a conflict
with the first 10 characters, but a conflict with the whole string is
pretty much discountable.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company