[PATCH] git-send-email: don't return undefined value in extract_valid_address()

[PATCH] git-send-email: don't return undefined value in extract_valid_address()

[Krzysztof Mazur]

In the fallback check, when Email::Valid is not available, the
extract_valid_address() does not check for success of matching regex,
and $1, which can be undefined, is always returned. Now if match
fails an empty string is returned.

Signed-off-by: Krzysztof Mazur <krzysiek@podlesie.net>
---
This fixes following warnings:
Use of uninitialized value in string eq at ./git-send-email.perl line 1017.
Use of uninitialized value in quotemeta at ./git-send-email.perl line 1017.
W: unable to extract a valid address from: x a.patch

when invalid email address was added by --cc-cmd,
./git-send-email.perl --dry-run --to a@podlesie.net --cc-cmd=echo x a.patch

 git-send-email.perl | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/git-send-email.perl b/git-send-email.perl
index 5a7c29d..045f25f 100755
--- a/git-send-email.perl
+++ b/git-send-email.perl
@@ -831,12 +831,12 @@ sub extract_valid_address {
 	$address =~ s/^\s*<(.*)>\s*$/$1/;
 	if ($have_email_valid) {
 		return scalar Email::Valid->address($address);
-	} else {
-		# less robust/correct than the monster regexp in Email::Valid,
-		# but still does a 99% job, and one less dependency
-		$address =~ /($local_part_regexp\@$domain_regexp)/;
-		return $1;
 	}
+
+	# less robust/correct than the monster regexp in Email::Valid,
+	# but still does a 99% job, and one less dependency
+	return $1 if $address =~ /($local_part_regexp\@$domain_regexp)/;
+	return "";
 }
 
 # Usually don't need to change anything below here.
-- 
1.8.0.283.gc57d856

Re: [PATCH] git-send-email: don't return undefined value in extract_valid_address()

[Junio C Hamano]

Krzysztof Mazur <krzysiek@podlesie.net> writes:

> In the fallback check, when Email::Valid is not available, the
> extract_valid_address() does not check for success of matching regex,
> and $1, which can be undefined, is always returned. Now if match
> fails an empty string is returned.

That much we can read from the code, but a bigger question is why
would it be a good thing for the callers?  Wouldn't they want to
be able to distinguish a failure from an empty string?

> Signed-off-by: Krzysztof Mazur <krzysiek@podlesie.net>
> ---
> This fixes following warnings:
> Use of uninitialized value in string eq at ./git-send-email.perl line 1017.
> Use of uninitialized value in quotemeta at ./git-send-email.perl line 1017.
> W: unable to extract a valid address from: x a.patch
>
> when invalid email address was added by --cc-cmd,
> ./git-send-email.perl --dry-run --to a@podlesie.net --cc-cmd=echo x a.patch

In other words, would we want to *hide* (not "fix") the warning?
Shouldn't we be barfing loudly and possibly erroring it out until
the user fixes her --cc-cmd?

>  git-send-email.perl | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/git-send-email.perl b/git-send-email.perl
> index 5a7c29d..045f25f 100755
> --- a/git-send-email.perl
> +++ b/git-send-email.perl
> @@ -831,12 +831,12 @@ sub extract_valid_address {
>  	$address =~ s/^\s*<(.*)>\s*$/$1/;
>  	if ($have_email_valid) {
>  		return scalar Email::Valid->address($address);
> -	} else {
> -		# less robust/correct than the monster regexp in Email::Valid,
> -		# but still does a 99% job, and one less dependency
> -		$address =~ /($local_part_regexp\@$domain_regexp)/;
> -		return $1;
>  	}
> +
> +	# less robust/correct than the monster regexp in Email::Valid,
> +	# but still does a 99% job, and one less dependency
> +	return $1 if $address =~ /($local_part_regexp\@$domain_regexp)/;
> +	return "";
>  }
>  
>  # Usually don't need to change anything below here.

Re: [PATCH] git-send-email: don't return undefined value in extract_valid_address()

[Krzysztof Mazur]

On Tue, Nov 20, 2012 at 12:27:36PM -0800, Junio C Hamano wrote:
> Krzysztof Mazur <krzysiek@podlesie.net> writes:
> 
> > In the fallback check, when Email::Valid is not available, the
> > extract_valid_address() does not check for success of matching regex,
> > and $1, which can be undefined, is always returned. Now if match
> > fails an empty string is returned.

Maybe the last line of comment should be changed to:

fails an empty string is returned to indicate failure.

> 
> That much we can read from the code, but a bigger question is why
> would it be a good thing for the callers?  Wouldn't they want to
> be able to distinguish a failure from an empty string?

In this case returning empty string does not make sense, so it's
really used to indicate failure.

> 
> > Signed-off-by: Krzysztof Mazur <krzysiek@podlesie.net>
> > ---
> > This fixes following warnings:
> > Use of uninitialized value in string eq at ./git-send-email.perl line 1017.
> > Use of uninitialized value in quotemeta at ./git-send-email.perl line 1017.
> > W: unable to extract a valid address from: x a.patch
> >
> > when invalid email address was added by --cc-cmd,
> > ./git-send-email.perl --dry-run --to a@podlesie.net --cc-cmd=echo x a.patch
> 
> In other words, would we want to *hide* (not "fix") the warning?
> Shouldn't we be barfing loudly and possibly erroring it out until
> the user fixes her --cc-cmd?
> 

Yes, it's just to hide the warning, the error (warning in this case) it's
already correctly generated:

W: unable to extract a valid address from: x a.patch

Maybe we should change it to an error?

Krzysiek

Re: [PATCH] git-send-email: don't return undefined value in extract_valid_address()

[Junio C Hamano]

Krzysztof Mazur <krzysiek@podlesie.net> writes:

> Yes, it's just to hide the warning, the error (warning in this case) it's
> already correctly generated:
>
> W: unable to extract a valid address from: x a.patch

But it is of no use if the message is sent out without the intended
recipient, no?  It is too late when you notice it.

> Maybe we should change it to an error?

At least, when we are not giving the "final sanity check [Y/n]?"
prompt, I think the code should error out.