Re: [RFC PATCHv2 1/2] am: add gitk patch format

[Chris Packham <judge.packham@gmail.com>]

On Sat, Sep 6, 2014 at 6:29 AM, Junio C Hamano <gitster@pobox.com> wrote:
> Chris Packham <judge.packham@gmail.com> writes:
>
>> So teaching git mailinfo to do s/^    // (either when asked to or
>> using some heuristic) would be a better approach? I also think we
>> should accept "Author:" as an acceptable fallback if "From:" is not
>> present.
>
> Not as "a fallback" in the sense that "Author:" should not be
> treated any specially when "am" (which stands for "apply mail") is\
> operating on the patches in e-mails.

I was proposing we avoid the "Patch does not have a valid e-mail
address." error when we can dwim and determine the email address from
"Author:". I originally was going to say "From:" should take
precedence but it would be another way to indicate that the true
author is not necessarily the person who sent the email.

> Whatever wants to convert the
> output from "log --pretty" as if it came from "log --pretty=email"
> would certainly need to flip "Author:" to "From:" (what should
> happen when it sees "From:" in the input, though???), and whatgver
> wannts to pick metainfo from "log --pretty" output like mailinfo
> does for "log --pretty=email" output would certainly need to pick
> the authorship from "Author:" (without paying any attention to
> "From:" if one exists).
>

Wow. I didn't know --pretty=email existed. Better yet it works for
diff-tree so gitk should probably be using that to produce something
that can be exported/imported easily.

I do wonder what the original use-case for "write commit to file" was.
Once it's been written to a file what is one supposed to do with it?
It's not something that 'git am' can consume (currently). Using 'git
apply' or 'patch' would lose the commit message plus you have to
manually stage/commit the changes.