Re: [PATCH v3 0/2] test: tests for the "double > from mailmap" bug

[Felipe Contreras <felipe.contreras@gmail.com>]

On Wed, Feb 15, 2012 at 12:49 AM, Junio C Hamano <gitster@pobox.com> wrote:
> Felipe Contreras <felipe.contreras@gmail.com> writes:
>
>> On Wed, Feb 15, 2012 at 12:18 AM, Junio C Hamano <gitster@pobox.com> wrote:
>>
>> And I don't understand why people want the obvious to be explained.
>
> Has it ever occurred to you the reason why people ask questions to you is
> perhaps because something that is obvious to you who wrote the patch is
> not obvious at all to others?  Has it also occurred to you that the
> majority of people who need to understand the patch during the review and
> 6 months down the road in "git log" output are not *you*?

Yes, that's why I am still listening. However I have not yet found a
question that cannot be answered from the simple description, except
the one you brought and I agreed to tackle.

I would like to think I have the capacity of empathy, so I would be
able to see if something cannot be inferred from the commit message.
Of course, I might be wrong, but so far the feedback has not been "no,
it's not obvious", but rather " yes, it's obvious... but...".

>> Your new point is "you can add a new thing that we did not have, but
>> it would not result in a good addition if that new thing is
>> irrelevant", but you already know what is the new thing from the
>> summary "'git blame -e' tests".
>
> It is not a "new point".  Jonathan, Peff and I all never said that it is
> unclear "what" your patch adds.  The suggestions for improvement given in
> this thread were all to explain "why" better.

I have heard both. And the "why" can be easily inferred, at least on
the first patch. The second one I yet have to fix, as I already
replied to you.

>> Everybody seems to assume that a simple commit message = bad. I disagree.
>
> If you find *everybody* seems to disagree with you, it would help to
> consider a slight possibility that you *might* be wrong.  And "simple" is
> not necessarily "sufficient and simple".

Of course, it's *possible*, but ad populum is not a valid argument.

There's many projects out there, and very few have as verbose commit
messages as git. I do not say they are doing it better, as many times
a lot of the verbose commit messages do help, but I don't think this
is the case.

Looking at the latest commits in the Linux kernel show good examples
of simple commit messages. Albeit they might be a bit "too simple", my
point that they do have a different view on what is "sufficient and
simple":

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=27c3afe6e1cf129faac90405121203962da08ff4
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=0d86f65ed0b727daa06d3aa176314cd175323db6
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=10f296cbfe3b93188c41463fd7a53808ebdbcbe3

Of course, what is "sufficient and simple" depends on a case by case
basis, but I wonder, if there's any case in which a single line in the
commit message summary would suffice, wouldn't adding missing
inconspicuous tests for something be it?

>> ... And I already pointed out the double standards.
>
> Sorry, but the absolute uniform standards do not exist, unless you are
> living in a fantasy land.  I expect better from list regulars as new
> contributors will inevitably learn from their behaviour (we also learn
> from our past mistakes).

Of course, but I can't help but wonder... Why so much fuss for simple
tests? And why you didn't bother to add tests for your new code as
well?

Cheers.

-- 
Felipe Contreras