From: Erick Mattos <erick.mattos@gmail.com>
Date: Sat, 31 Oct 2009 01:42:47 GMT
Subject: Re: [PATCH] Changed timestamp behavior of options -c/-C/--amend
Message-ID: <55bacdd30910301842m77cf5d7fi8389e305fd439a72@mail.gmail.com>
In-Reply-To: <7vocnoe83m.fsf@alter.siamese.dyndns.org>

You are completely right.

All your concern is relevant and the whole problem must be re-engineered.

The good news is that I have almost finished it and I will be starting
a new thread with the new solution in a few minutes.

Regards

2009/10/30 Junio C Hamano <gitster@pobox.com>:
> Erick Mattos <erick.mattos@gmail.com> writes:
>
>> I don't see a use for comparing the author and committer because I can
>> use as template my own commits or others'.
>
> You _can_ use whichever irrelevant commit as a template, but "you _can_"
> is different from what it means, and what is and what is not _sensible_.
>
> You may be rewriting somebody else's patch (e.g. fixing up a typo in the
> message, or changing the implementation, or both).  If you are going to
> keep the authorship, you are saying that "it is still _his_ code, not
> mine".  In such a case, it never makes sense to change the timestamp, if
> that author is somebody other than you.  After all that other guy may not
> even be aware of what you are doing when you make this commit; he may be
> in bed sound asleep in a different timezone.
>
> In another scenario, if your fix-up is very significant, even if you
> started from somebody else's patch, you may want to say "now this is my
> patch, the original author may have given me some inspiration, but the
> changes in this commit, including all the bugs, are mine".  The same
> applies if you looked at the problem description of somebody' patch, and
> did your own solution without using anything from his commit.
>
> At that point, you would want the resulting commit to say it was written
> by you at this moment.  You do not want to see -c/-C/--amend to retain any
> part of the authorship (not just timestamp) from the original commit.
>
>    Side note. You may be fixing your own patch, in which case you may or
>    may not consider your change significant, but at the time of either
>    old timestamp or current time, you were working on this change, so
>    using the current timestamp instead of using the old one is not a big
>    deal, and that is why I think committer==author may be a good
>    heuristic when deciding to touch or not touch the timestamp.
>
>    But in general I do not like such dwim that depends on who you are (it
>    makes it harder to explain, even if the end result may be useful in
>    practice), so I'd rather not to see such a code for this topic if we
>    can avoid it.
>
> In short, I do not think it makes sense to change only the timestamp while
> keeping the author.  The issue is not "timestamp behaviour" with "use new
> timestamp" option, but rather is an ability to declare "Now this is a
> commit made _by me_ and _now_; iow, I take authorship for this change",
> even when you reuse the commit log message from somewhere else.
>
> So what is needed is an option to tell -c/-C/--amend to reuse _only_ the
> message but no authorship information from the original commit, I think.
>