[PATCH] git-submodule: fix expansion of depth for cmd_update

[PATCH] git-submodule: fix expansion of depth for cmd_update

[Jacob Keller]

From: Jacob Keller <jacob.keller@gmail.com>

The depth variable already contains "--depth=", so expanding it with an
additional --depth when invoking the update-clone git submodule--helper
is incorrect.

Signed-off-by: Jacob Keller <jacob.keller@gmail.com>
---

I'm *reasonably* sure this is correct, but I am not sure how to test it.
It's possible that it expands to "--depth --depth=N" and somehow this gets
handled properly?

 git-submodule.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/git-submodule.sh b/git-submodule.sh
index c7f58c5756f7..4e7fc8bf3652 100755
--- a/git-submodule.sh
+++ b/git-submodule.sh
@@ -547,7 +547,7 @@ cmd_update()
 		${update:+--update "$update"} \
 		${reference:+"$reference"} \
 		${dissociate:+"--dissociate"} \
-		${depth:+--depth "$depth"} \
+		${depth:+"$depth"} \
 		$recommend_shallow \
 		$jobs \
 		-- \
-- 
2.23.0.245.gf157bbb9169d

Re: [PATCH] git-submodule: fix expansion of depth for cmd_update

[Junio C Hamano]

Jacob Keller <jacob.e.keller@intel.com> writes:

> From: Jacob Keller <jacob.keller@gmail.com>
>
> The depth variable already contains "--depth=", so expanding it with an
> additional --depth when invoking the update-clone git submodule--helper
> is incorrect.
>
> Signed-off-by: Jacob Keller <jacob.keller@gmail.com>
> ---
>
> I'm *reasonably* sure this is correct, but I am not sure how to test it.
> It's possible that it expands to "--depth --depth=N" and somehow this gets
> handled properly?

I agree with your eyeballing of all the assignments to the variable,
and other references to $depth take either one of these two forms:

	git submodule--helper ... ${depth:+"$depth"} ...
	git submodule--helper ... $depth ...

As long as "git submodule ... --depth <depth> ..." gets called with
<depth> that does not have $IFS, either would work fine, but the
former is correct even when <depth> has problematic characters in it
and your patch uses that form, too).

However.

The command line parser for update_clone() stuffs --depth as a
string to suc.depth, and then the machinery ends up calling 
prepare_to_clone_next_submodule() with such an instance of suc
(struct submodule_update_clone).  Then that function just pushes the
suc->depth to an argv array used to spawn a "submodule--helper clone".

So passing "--depth --depth=23" would be "correct", sadly, in that
codepath (I am not saying other codepaths would not call the same
prepare_to_clone_next_submodule() with "--depth 23", as I didn't
check, and if there is such a codepath, it would break).  

We may need to clean the mess up X-<.

> diff --git a/git-submodule.sh b/git-submodule.sh
> index c7f58c5756f7..4e7fc8bf3652 100755
> --- a/git-submodule.sh
> +++ b/git-submodule.sh
> @@ -547,7 +547,7 @@ cmd_update()
>  		${update:+--update "$update"} \
>  		${reference:+"$reference"} \
>  		${dissociate:+"--dissociate"} \
> -		${depth:+--depth "$depth"} \
> +		${depth:+"$depth"} \
>  		$recommend_shallow \
>  		$jobs \
>  		-- \

RE: [PATCH] git-submodule: fix expansion of depth for cmd_update

[Keller, Jacob E]

> -----Original Message-----
> From: git-owner@vger.kernel.org [mailto:git-owner@vger.kernel.org] On Behalf Of
> Junio C Hamano
> Sent: Thursday, August 22, 2019 2:01 PM
> To: Keller, Jacob E <jacob.e.keller@intel.com>
> Cc: git@vger.kernel.org; Jacob Keller <jacob.keller@gmail.com>
> Subject: Re: [PATCH] git-submodule: fix expansion of depth for cmd_update
> 
> Jacob Keller <jacob.e.keller@intel.com> writes:
> 
> > From: Jacob Keller <jacob.keller@gmail.com>
> >
> > The depth variable already contains "--depth=", so expanding it with an
> > additional --depth when invoking the update-clone git submodule--helper
> > is incorrect.
> >
> > Signed-off-by: Jacob Keller <jacob.keller@gmail.com>
> > ---
> >
> > I'm *reasonably* sure this is correct, but I am not sure how to test it.
> > It's possible that it expands to "--depth --depth=N" and somehow this gets
> > handled properly?
> 
> I agree with your eyeballing of all the assignments to the variable,
> and other references to $depth take either one of these two forms:
> 
> 	git submodule--helper ... ${depth:+"$depth"} ...
> 	git submodule--helper ... $depth ...
> 
> As long as "git submodule ... --depth <depth> ..." gets called with
> <depth> that does not have $IFS, either would work fine, but the
> former is correct even when <depth> has problematic characters in it
> and your patch uses that form, too).
> 
> However.
> 
> The command line parser for update_clone() stuffs --depth as a
> string to suc.depth, and then the machinery ends up calling
> prepare_to_clone_next_submodule() with such an instance of suc
> (struct submodule_update_clone).  Then that function just pushes the
> suc->depth to an argv array used to spawn a "submodule--helper clone".
> 
> So passing "--depth --depth=23" would be "correct", sadly, in that
> codepath (I am not saying other codepaths would not call the same
> prepare_to_clone_next_submodule() with "--depth 23", as I didn't
> check, and if there is such a codepath, it would break).
> 

Ok, so it's technically correct, but weird. I had trouble understanding it.

> We may need to clean the mess up X-<.
> 

Yea, I got confused looking at this code. Though if I recall, we were working towards rewriting it in C anyways (which is what the submodule--helper is doing).

Thanks,
Jake

> > diff --git a/git-submodule.sh b/git-submodule.sh
> > index c7f58c5756f7..4e7fc8bf3652 100755
> > --- a/git-submodule.sh
> > +++ b/git-submodule.sh
> > @@ -547,7 +547,7 @@ cmd_update()
> >  		${update:+--update "$update"} \
> >  		${reference:+"$reference"} \
> >  		${dissociate:+"--dissociate"} \
> > -		${depth:+--depth "$depth"} \
> > +		${depth:+"$depth"} \
> >  		$recommend_shallow \
> >  		$jobs \
> >  		-- \