Ensuring commit messages with correct copyright info

Ensuring commit messages with correct copyright info

[Lars Magne Ingebrigtsen]

The thing that comes up, again and again, as the main reason we should
be able to edit commit messages (and therefore we should have
ChangeLogs, because we can edit those) is that we (er, mainly me :-))
sometimes forget to put

Copyright-paperwork-exempt: yes

into the commit messages.

So why not concentrate of fixing that particular problem?  I think
something that might help is that if the committer is different from the
author, then there should always be something in the commit message that
explicitly says what the copyright status of the work is.

That is, unless the line

Copyright-paperwork-exempt: yes

or the line

Copyright-assigned: yes

is present in the commit message, then the commit... hooks? (the thing
that refuses to check in files with trailing newlines etc; I don't know
the details about how that works) should refuse the commit.

Then it should be pretty impossible to forget this stuff, and there's
one less reason to want to edit the commit messages.

(Note: This hook should only be run if there's an Author: header in the
vc log buffer.  Everybody who has commit rights have already assigned
copyright, so it's unnecessary to add those if the committer and the
author are the same person.)

(Copyright-disclaimed: yes should also be a possible value, I guess.)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Ensuring commit messages with correct copyright info

[Lars Magne Ingebrigtsen]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Then it should be pretty impossible to forget this stuff, and there's
> one less reason to want to edit the commit messages.

(And if somebody still manages to get it wrong, we should have a rule
saying that commits with wrong metadata should be reverted and reapplied
with the correct data.  If this is done within a reasonable time, that's
usually pretty easy to do.)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Ensuring commit messages with correct copyright info

[Eli Zaretskii]

> From: Lars Magne Ingebrigtsen <larsi@gnus.org>
> Date: Tue, 08 Mar 2016 19:47:29 +0100
> 
> The thing that comes up, again and again, as the main reason we should
> be able to edit commit messages (and therefore we should have
> ChangeLogs, because we can edit those) is that we (er, mainly me :-))
> sometimes forget to put
> 
> Copyright-paperwork-exempt: yes
> 
> into the commit messages.

It's one problem, but it's not the only one.  Just look at the
generated ChangeLog file, and you will see the blunders.  We need to
educate contributors to do a better job, and I see no way of doing
that without some means of fixing the mistakes.

> So why not concentrate of fixing that particular problem?  I think
> something that might help is that if the committer is different from the
> author, then there should always be something in the commit message that
> explicitly says what the copyright status of the work is.
> 
> That is, unless the line
> 
> Copyright-paperwork-exempt: yes
> 
> or the line
> 
> Copyright-assigned: yes
> 
> is present in the commit message, then the commit... hooks? (the thing
> that refuses to check in files with trailing newlines etc; I don't know
> the details about how that works) should refuse the commit.
> 
> Then it should be pretty impossible to forget this stuff, and there's
> one less reason to want to edit the commit messages.

So you are suggesting that instead of doing TRT in perhaps 2% of
commit messages, we should now add a completely redundant text in 98%
of them?  How does this make sense?

> (Note: This hook should only be run if there's an Author: header in the
> vc log buffer.  Everybody who has commit rights have already assigned
> copyright, so it's unnecessary to add those if the committer and the
> author are the same person.)

I already use --author in those cases, why should I do anything else?

Re: Ensuring commit messages with correct copyright info

[Lars Magne Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> So you are suggesting that instead of doing TRT in perhaps 2% of
> commit messages, we should now add a completely redundant text in 98%
> of them?  How does this make sense?

98%?  My guess would be that the number of commits that have an Author:
header is about 10% of the total volume.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Ensuring commit messages with correct copyright info

[Eli Zaretskii]

> From: Lars Magne Ingebrigtsen <larsi@gnus.org>
> Cc: emacs-devel@gnu.org
> Date: Tue, 08 Mar 2016 20:25:06 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > So you are suggesting that instead of doing TRT in perhaps 2% of
> > commit messages, we should now add a completely redundant text in 98%
> > of them?  How does this make sense?
> 
> 98%?  My guess would be that the number of commits that have an Author:
> header is about 10% of the total volume.

OK, 90%.

Instead, how about setting up a secure server on fencepost.gnu.org
that Emacs could automatically query, and insert the necessary
Copyright-paperwork-exempt header if needed?  That'd be really
helpful.

Re: Ensuring commit messages with correct copyright info

[Lars Magne Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> 98%?  My guess would be that the number of commits that have an Author:
>> header is about 10% of the total volume.
>
> OK, 90%.

No, 10%.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Ensuring commit messages with correct copyright info

[John Wiegley]

>>>>> Eli Zaretskii <eliz@gnu.org> writes:

> Instead, how about setting up a secure server on fencepost.gnu.org that
> Emacs could automatically query, and insert the necessary
> Copyright-paperwork-exempt header if needed? That'd be really helpful.

This is a great idea, Eli, I'll see what I can do. Just using TRAMP and
querying the file shouldn't be too hard.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2