Re: [ANNOUNCEMENT] GDB 8.1 release branch created!

[Yao Qi <qiyaoltc at gmail dot com>]

On Thu, Feb 1, 2018 at 5:34 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Yao Qi <qiyaoltc@gmail.com>
>> Date: Thu, 1 Feb 2018 16:51:36 +0000
>> Cc: Simon Marchi <simon.marchi@ericsson.com>, GDB Patches <gdb-patches@sourceware.org>
>>
>> On Thu, Feb 1, 2018 at 4:27 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>> > Maybe I'm missing something, but isn't the commit log entry supposed
>> > to have the same text as the ChangeLog for that change?
>>
>> My understanding is that ChangeLog entry should be copied at the end
>> of commit log
>
> I believe I did that, except that I moved the motivation for the
> change before the ChangeLog header, as it belongs to the preamble of
> the Git log.  (In ChangeLog, there's no subdivision of the entry.)
>

It is absolutely fine to put the reason of the change in the commit
log, but it is questionable to put commit log into ChangeLog.  At
least, nowadays, "commit log" is more than ChangeLog entry.  This
is the convention we followed for several years.

>> > And what's wrong with having error messages in ChangeLog's, anyway?
>>
>> Each entry in ChangeLog describes the changes to the code,
>> https://www.gnu.org/prep/standards/standards.html#Change-Log-Concepts
>> IMO, error messages are out of the scope of what ChangeLog is intended
>> to describe.
>
> This is changing, and some GNU projects already do it differently.  I
> believe the next version of standards.texi will change the
> recommendations to be more lax.  And using different style for each
> project is a pain.  Is GDB really so rigid as to not tolerate this?
>

GDB is evolving, so I am completely fine to changes, but you
made the change of the way of writing changelog, which is not
discussed before.

> Please note that the error message I cited is just part of the
> motivation for the change, so we are really talking about including
> the motivation in the ChangeLog.
>

Yes, do we want to change the rule of writing ChangeLog, so that
we can/should put the motivation for the change?  Is the motivation
mandatory in the changelog or optional?

>> Also, I've never seen such ChangeLog entry (with error messages)
>> before.
>
> That's not necessarily a reason for rejecting such entries.
>
> So: what is the project's take on this, if there is one?

We need a clear rule on writing changelog, otherwise how do we
write changelog or review patches?

-- 
Yao (齐尧)