Re: [PATCH] git-add--interactive.perl: Permit word-based diff

[Mark Abraham <mark.j.abraham@gmail.com>]

On Tue, Jun 18, 2013 at 7:23 PM, Jeff King <peff@peff.net> wrote:
> On Tue, Jun 18, 2013 at 09:22:16AM +0200, Thomas Rast wrote:
>
>> [I don't seem to have received a copy of the original mail, so I can
>> only guess...]
>
> Yes, the original doesn't seem to have made it to the list. Sorry, I
> don't have a copy (I am in the habit of deleting direct mails that are
> cc'd to the list, as I keep a separate list archive).
>
> Mark, did you happen to send an HTML mail? The list will silently
> reject such mail.

Yes, that was probably it. I tried to find a gmail configuration, but
I now discover it is done per-email, not globally. Apologies. I have
forwarded the original to Thomas, but based on current feedback, it
seems not worth re-sending the original mail to the list. See below.

>> > Note that the number of lines in your --word-diff=color hunk and the
>> > actual diff will not necessarily be the same.  What happens if I split a
>> > hunk with your patch?
>>
>> If it's actually what you hint at, there's another problem: the word
>> diff might not even have the same number of hunks.  For example, a
>> long-standing bug (or feature, depending on POV) of word-diff is that it
>> does not take opportunities to completely drop hunks that did not make
>> any word-level changes.
>
> Yeah, I didn't even think of that.
>
> In general, I think one can assume 1-to-1 correspondence between whole
> regular diffs and whole word-diffs, but not below that (i.e., neither a
> correspondence between hunks nor between lines).
>
> With something like contrib/diff-highlight, you get some decoration, and
> can assume a 1-to-1 line correspondence.

My choice of "permit" in the description was not best. My
implementation showed a word-based diff, but preserved the existing
mechanism for actually applying the hunk. I understand the way
colorization in git-add--interactive.perl works right now is to
colorize one version to display and use another - I think I preserved
that. I intended to permit the user to choose to show a word-based
diff of a patch during interactive add.

> However, I think that when reviewing text (especially re-wrapped
> paragraphs) that word-diff can be much easier to read, _because_ it
> throws away the line correspondence. To be correct, though, I think we
> would have to simply show the whole word-diff for a file and say "is
> this OK?". Which sort of defeats the purpose of "add -p" as a hunk
> selector (you could just as easily run "git diff --color-words foo" and

Hmm, I will have to re-consider the implications on that kind of
workflow. Thanks!

> "git add foo"). But it does save keystrokes if your workflow is to
> simply "add -p" everything to double-check it before adding.

Yes, that was what I was aiming to make easier.

> So I dunno. I could see myself using it, but I certainly wouldn't want a
> config variable that turns it on all the time (which is what the
> original patch did).

Good point. What I think I really want is "git add
--interactive=color" (with or without --patch) to permit the user to
choose to see the (colorized) word-based diff when they want one. I
now see that the config file approach in my proposed patch doesn't go
close enough to that to be worth considering further.

I think a proper implementation of the above command would have to
* add something to builtin_add_options in builtin/add.c,
* set a new static variable in add.c, and
* extend the calling logic for interactive_add() and/or
run_add_interactive() accordingly,
so that the perl script can get the user's choice on the command line
and not from a config file. And only respond when colorization is
configured.

Does --patch=color, --interactive=color or adding new option
--color-words make more sense?

I'll have a think about that and get back to you guys.

Thanks!

Mark

> -Peff