Re: Is it time to drop ChangeLogs?

[Nikolaus Rath]

On Mar 07 2016, Eli Zaretskii <address@hidden> wrote:
>> From: Nikolaus Rath <address@hidden>
>> Cc: address@hidden
>> Date: Mon, 07 Mar 2016 13:15:16 -0800
>> 
>> On Mar 07 2016, Eli Zaretskii <address@hidden> wrote:
>> >> From: Nikolaus Rath <address@hidden>
>> >> Date: Mon, 07 Mar 2016 12:46:30 -0800
>> >> 
>> >> When I submitted my first Emacs patch, I was astonished when I was asked
>> >> to re-submit with my commit message essentially duplicated in the
>> >> ChangeLog.
>> >
>> > One is just a copy of the other, so I fail to see a problem, or a
>> > reason for astonishment.
>> 
>> One just being a copy of the other is the reason for astonishment.
>
> I still don't see why.  It's a simple request that is easily
> accomplished.  I mentioned several other GNU projects that do the
> same.  It's accepted practice.

Asking me to walk three times around my chair before making the commit
is also a simple request that's easily accomplished. That doesn't make
it any less astonishing. There is just no apparent good reason for this
duplication. Thus it astonishes.

>> >> How can it be an increased burden if reviewers have to review just one
>> >> thing (the commit message) instead of two (commit message and
>> >> ChangeLog)?
>> >
>> > No one reviews the same text twice, so doing this will not affect the
>> > review directly.
>> 
>> Well, you are the one who claimed that it makes a difference.
>
> No, you've misunderstood what I wrote.
>
>> > Indirectly, it will make sure your patches are cleaner, because
>> > summarizing what you did will frequently reveal subtle blunders and
>> > things you forgot.
>> 
>> We are talking about the advantages of making a copy of that summary,
>> after it has been written.
>
> You didn't read what I wrote about the current system.  If we stop
> producing ChangeLog files, there will be no reason for having detailed
> enough commit log messages, and soon enough there will be no
> summaries.

I don't see how one follows from the other. Why isn't whatever measure
you are currently take to ensure detailed ChangeLog files equally
suitable to ensure detailed commit messages?


Best,
-Nikolaus

-- 
GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F
Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F

             »Time flies like an arrow, fruit flies like a Banana.«