git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jon Seymour <jon.seymour@gmail.com>
To: Brian Gernhardt <brian@gernhardtsoftware.com>
Cc: robbat2@gentoo.org, git@vger.kernel.org
Subject: Re: [PATCH v1] stash show: fix breakage in 1.7.3
Date: Sat, 25 Sep 2010 16:19:50 +1000	[thread overview]
Message-ID: <AANLkTimK4nTr9bS0L9xc9mGo5F900pUAOQoDnZ9HMdEz@mail.gmail.com> (raw)
In-Reply-To: <AF7816CF-7515-4386-9736-417005434433@gernhardtsoftware.com>

On Sat, Sep 25, 2010 at 2:45 PM, Brian Gernhardt
<brian@gernhardtsoftware.com> wrote:
>
> On Sep 24, 2010, at 11:32 PM, Jon Seymour wrote:
>
>> due to a faulty assumption that:
>>   git rev-parse --no-revs -- stash@{0}
>
> This assumption is faulty, it should be "git rev-parse --no-revs --flags stash@{0}", which works properly for all revision arguments and flags _except_ -q and --quiet.

Agreed. If I recall the evolution of the code, I originally had
--no-revs --flags, but then found -q was being eaten. The only way to
prevent it being eaten was to remove --flags and add --, which "fixed"
the -q problem but created the git stash show xxx problem. Of course,
had my unit tests been more thorough, I would have picked this, but
alas they were not.

>
>> This revision further simplifies the parsing code
>> by removing use of git rev-parse for FLAGS parsing
>> altogether.
>
> That is simpler, and does fix this specific issue.  However, I would strongly argue that "git rev-parse --no-revs --flags" is broken.  I really don't have the time tonight or probably this weekend to work on it, but git-rev-parse should only take "-q" and "--quiet" for itself if "--verify" was passed.  (Since that is the only mode in which rev-parse uses quiet, AFAIK.)
>

This seems like a reasonable compromise to me.

> Possibly rev-parse should also (or instead) separate "arguments for rev-parse" and "arguments rev-parse is parsing" using the standard "--".  I don't know if this will affect any current users.
>

I suspect this could be problematic.

Current behaviour:

   $ git rev-parse --flags -X -- Y -Z
   -X

Here -- is being used to say, don't interpret anything after -- as
being flag-like even if it could be flag-like.

So, if a command foo uses git rev-parse to parse its own arguments:

    foo -X -- Y -Z

then I suspect, it would only want to treat -X as being flag-like,
from the user of foo's point of view, both Y and -Z are not meant to
be interpreted by foo.

It might make sense for --flags to be an instruction to git rev-parse
not to interpret any subsequent arguments as git rev-parse options so
that:

  $ git rev-parse --flags -q --no-flags -- --revs-only

would output:

  -q --no-flags

That is, prevent -q being eaten by git rev-parse, but preserve
existing interpretation of --.

> ~~ Brian
>
>

  reply	other threads:[~2010-09-25  6:22 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-09-24 19:19 git-1.7.3 breakage: "git stash show xxx" doesn't show anything Robin H. Johnson
2010-09-24 20:01 ` Brandon Casey
2010-09-24 20:27   ` Brian Gernhardt
2010-09-24 20:40     ` [PATCH] t/t3903-stash: improve testing of git-stash show Brandon Casey
2010-09-24 20:43       ` Brian Gernhardt
2010-09-24 20:50         ` Brandon Casey
2010-09-24 21:49           ` Brian Gernhardt
2010-09-24 22:11             ` [PATCH] git-stash: fix flag parsing Brian Gernhardt
2010-09-24 22:15             ` Brian Gernhardt
2010-09-25  2:58               ` Jon Seymour
2010-09-27  4:36               ` Junio C Hamano
2010-09-27 15:32                 ` [PATCH] stash: simplify parsing fixes Jon Seymour
2010-09-25  2:54 ` [PATCH] stash show: fix breakage in 1.7.3 Jon Seymour
2010-09-25  3:32 ` [PATCH v1] " Jon Seymour
2010-09-25  4:45   ` Brian Gernhardt
2010-09-25  6:19     ` Jon Seymour [this message]
2010-09-25  7:15   ` Jon Seymour
2010-09-27 15:38   ` Jon Seymour
2010-09-25  7:19 ` [PATCH/RFC] rev-parse: stop interpreting flags as options to rev-parse once --flags is specified Jon Seymour
2010-09-25  7:19   ` Jon Seymour
2010-09-25  7:25     ` Jon Seymour
2010-09-26  7:11       ` Junio C Hamano
2010-09-26 14:39         ` Jon Seymour
2010-09-26 14:44         ` [PATCH v7 0/3] rev-parse: allow --flags to output rev-parse-like flags Jon Seymour
2010-09-26 14:44         ` [PATCH v7 1/3] rev-parse: update Documentation of --flags Jon Seymour
2010-09-26 14:44         ` [PATCH v7 2/3] rev-parse: add tests for git rev-parse --flags Jon Seymour
2010-09-26 14:44         ` [PATCH v7 3/3] rev-parse: stop interpreting flags as options to rev-parse once --flags is specified Jon Seymour

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=AANLkTimK4nTr9bS0L9xc9mGo5F900pUAOQoDnZ9HMdEz@mail.gmail.com \
    --to=jon.seymour@gmail.com \
    --cc=brian@gernhardtsoftware.com \
    --cc=git@vger.kernel.org \
    --cc=robbat2@gentoo.org \
    /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).