git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: Stefan Beller <stefanbeller@googlemail.com>
Cc: git@vger.kernel.org, jrnieder@gmail.com, trast@student.ethz.ch
Subject: Re: [PATCH 0/2] git diff -q option removal
Date: Wed, 17 Jul 2013 10:04:08 -0700	[thread overview]
Message-ID: <7vvc496ruf.fsf@alter.siamese.dyndns.org> (raw)
In-Reply-To: <1373970487-32595-1-git-send-email-stefanbeller@googlemail.com> (Stefan Beller's message of "Tue, 16 Jul 2013 12:28:05 +0200")

Stefan Beller <stefanbeller@googlemail.com> writes:

> The changes in the following patch are in diff_no_index.c, but the
> diff_no_index(...) is called from cmd_diff, which is in builtin/diff.c
> That cmd_diff is actually called from git.c having the
> { "diff", cmd_diff }, entry in handle_internal_command.
>
> My question now is this: Why is the builtin/diff.c relying on stuff
> outside of builtin/ ? Wouldn't it be better to move all these files
> (such as diff_no_index.c) into the builtin folder as well?

Builtins link all sorts of stuff from outside, e.g. diff.c and
diffcore-*.c at the toplevel.  I do not see diff_no_index.c is any
different, so I am probably not understanding your question.

> Regarding the removal of the -q option, I tried it in the second patch.
> Is it as easy as that, or am I missing the point?
> 
> The first patch doesn't change the behavior, so I'd assume it's safe to 
> apply it to origin/sb/misc-fixes, whereas the second patch will make 
> git diff complain about the -q option, so I'd assume it would wait for the
> next major release?
>
> Before:
> 	touch actual_file
> 	git diff -q  actual_file no_file
> 	error: Could not access 'no_file'

Hmm, do you really get that error message?  I think you would get

    fatal: ambiguous argument 'no_file': unknown revision or path not in the working tree.

> 	echo $?
> 	1

The command line parsing infrastructure has changed vastly since
"show-diff" days (see below for a history lesson); I think your
"Before" should read more like this

	git diff -q -- actual_file no_file

and it should not show removal of no_file in its output.  E.g. in
git.git

	$ git reset --hard
        $ rm COPYING
        $ git diff -q -- COPYING

should show nothing.

I personally think "-q" no longer makes sense in today's codebase,
but I am not convinced that removal of '-q' from the proper "git
diff-files" and the "git diff --no-index" (aka "I am too lazy to
teach our diff enhancement to other people's diff implementations,
so let's throw in a "files do not have to be tracked in Git
repository at all" mode") is the right direction to go.

The "-q" option is a remnant from the "show-diff" command, the
precursor of today's "git diff-files" (back then, we didn't even
have "git" potty.  The user literally typed "show-diff", not "git
show-diff").

ca2a0798 ([PATCH] Add "-q" option to show-diff.c, 2005-04-15) added
that option.  Back then, we did not have pathspec matching, and we
iterated over command line arguments, and required all of them exist
as filesystem entities.  "-q" was a way to defeat that "you name a
file, it must exist in the working tree" safety, and also at the
same time not give output for such a file that was removed from the
working tree.

These days, the former "safety" is enforced by the generalized
revision parser ("is it a path or is it a rev?") code and the "--"
delimiter on the command line is the way to defeat it.  The latter
is done by giving a filtering specification that lack D to the
"--diff-filter".

If we wanted to make "-q" follow the spirit of its original addition
to "show-diff" again, we could internally add a diff-filter when the
"-q" option is parsed.

"git diff -q ..." is "git diff --diff-filter=ACMRTUB ...", and "git
diff -q --diff-filter=AD" is "git diff --diff-filter=A".  That would
let us remove the special case for SILENT_ON_REMOVED, and also make
"-q" work across various commands in the "diff" family.  It might
even make it work for "diff --no-index", but I didn't bother to
check.

> After:
> 	touch actual_file
> 	git diff -q  actual_file no_file
> 	fatal: invalid diff option/value: -q
> 	echo $?
> 	128
>
> Thanks,
> Stefan
>
> Stefan Beller (2):
>   diff --no-index: remove nonfunctional "-q" handling
>   git diff: Remove -q option to stay silent on missing files.
>
>  Documentation/git-diff-files.txt | 6 +-----
>  diff-no-index.c                  | 5 -----
>  2 files changed, 1 insertion(+), 10 deletions(-)

  parent reply	other threads:[~2013-07-17 17:04 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-07-14 21:35 [PATCH 1/4] daemon.c:handle: Remove unneeded check for null pointer Stefan Beller
2013-07-14 21:35 ` [PATCH 2/4] commit: Fix a memory leak in determine_author_info Stefan Beller
2013-07-14 21:49   ` Jonathan Nieder
2013-07-14 21:35 ` [PATCH 3/4] diff-no-index: Remove unused variable Stefan Beller
2013-07-14 22:07   ` Jonathan Nieder
2013-07-16 10:28     ` [PATCH 0/2] git diff -q option removal Stefan Beller
2013-07-16 10:28       ` [PATCH 1/2] diff --no-index: remove nonfunctional "-q" handling Stefan Beller
2013-07-16 10:28       ` [PATCH 2/2] diff: Remove -q to stay silent on missing files Stefan Beller
2013-07-17 17:04       ` Junio C Hamano [this message]
2013-07-17 18:06         ` [PATCH 0/2] git diff -q option removal Junio C Hamano
2013-07-17 20:05         ` Stefan Beller
2013-07-18  0:30         ` [PATCH 0/6] Deprecating "diff-files -q" Junio C Hamano
2013-07-18  0:30           ` [PATCH 1/6] diff: pass the whole diff_options to diffcore_apply_filter() Junio C Hamano
2013-07-18  0:30           ` [PATCH 2/6] diff: factor out match_filter() Junio C Hamano
2013-07-18  0:30           ` [PATCH 3/6] diff: preparse --diff-filter string argument Junio C Hamano
2013-07-18  0:30           ` [PATCH 4/6] diff: reject unknown change class given to --diff-filter Junio C Hamano
2013-07-18  0:30           ` [PATCH 5/6] diff: allow lowercase letter to specify what change class to exclude Junio C Hamano
2013-07-18  0:30           ` [PATCH 6/6] diff: deprecate -q option to diff-files Junio C Hamano
2013-07-19  3:20             ` [PATCH 7/6] diff: remove "diff-files -q" at Git 2.0 version boundary Junio C Hamano
2013-07-19  3:31             ` [PATCH 6/6] diff: deprecate -q option to diff-files Jonathan Nieder
2013-07-19  7:24               ` Junio C Hamano
2013-07-19 21:01                 ` Jonathan Nieder
2013-07-14 21:35 ` [PATCH 4/4] diff.c: Do not initialize a variable, which gets reassigned anyway Stefan Beller
2013-07-14 22:18   ` Jonathan Nieder
2013-07-14 21:48 ` [PATCH 1/4] daemon.c:handle: Remove unneeded check for null pointer Jonathan Nieder

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=7vvc496ruf.fsf@alter.siamese.dyndns.org \
    --to=gitster@pobox.com \
    --cc=git@vger.kernel.org \
    --cc=jrnieder@gmail.com \
    --cc=stefanbeller@googlemail.com \
    --cc=trast@student.ethz.ch \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).