Re: A little report on informal commit tag usage

From: Michael Paquier <michael(at)paquier(dot)xyz>
To: Thomas Munro <thomas(dot)munro(at)gmail(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: A little report on informal commit tag usage
Date: 2019-07-16 07:43:36
Message-ID: 20190716074336.GG1439@paquier.xyz
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Jul 15, 2019 at 05:49:26PM +1200, Thomas Munro wrote:
> I would have tried to exclude the first line messages if I'd thought
> of that. But anyway, the reason for the low Doc number is case
> sensitivity. I ran that on a Mac and its lame collation support failed
> me in the "sort" step (also -i didn't do what I wanted, but that
> wasn't the issue). Trying again on FreeBSD box and explicitly setting
> LANG for the benefit of anyone else wanting to run this (see end), and
> then removing a few obvious false matches, I now get similar numbers
> in most fields but a higher "doc" number:
>
> 767 Author
> 9 Authors
> 144 Backpatch-through
> 55 Backpatch
> 14 Bug
> 14 Co-authored-by
> 27 Diagnosed-by
> 1599 Discussion
> 119 doc
> 36 docs
> 284 Reported-by
> 5 Review
> 8 Reviewed by
> 460 Reviewed-by
> 7 Security
> 9 Tested-by

Thanks for those numbers. I am wondering if we could do a bit of
consolidation here and write a page about this stuff on the wiki.
Getting the "Discussion" field most of the time is really cool.

I think that we could get some improvements on the following things.
Here is a set of ideas:
- Avoid "Authors" and replace it with "Author" even if there are
multiple authors.
- Avoid having multiple entries for each one of them? For example we
have a couple of commits listed listing one "Reviewed-by" field with
one single name.
- Most commit entries to not use the email address with the name of
the author, reviewer, tester or reporter. Perhaps we should give up
on that?
- Keep "Backpatch-through", not "Backpatch".
- Keep "Reviewed-by", not "Reviewed by" nor "Review".

"Security" is a special case, we append it to all the CVE-related
commits.

That is mainly a matter of taste, but I tend to prefer the following
format, protecting usually the order:
- Diagnosed-by
- Author
- Reviewed-by
- Discussion
- Backpatch-through
- I tend to have only one "Reviewed-by" entry with a list of names,
same for "Author" and "Reported-by".
- Only names, no emails.

As mentioned on different threads, "Discussion" is the only one we had
a strong agreement with. Could it be possible to consider things like
Author, Reported-by, Reviewed-by or Backpatch-through for example and
extend to that? The first three ones are useful for parsing the
commit logs. The fourth one is handy so as there is no need to look
at a full log tree with git log --graph or such, which is something I
do from time to time to guess down to where a fix has been applied (I
tend to avoid git_changelog).
--
Michael

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Fabien COELHO 2019-07-16 07:58:46 Re: pgbench - extend initialization phase control
Previous Message Rajkumar Raghuwanshi 2019-07-16 07:37:45 getting ERROR "relation 16401 has no triggers" with partition foreign key alter