Re: [PATCH 3/3] Remove stray error message from sed

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 1, 2010, 17:44 UTC
Message-ID: <7veigqvea6.fsf@alter.siamese.dyndns.org>
In-Reply-To: <1275384022-12131-4-git-send-email-artagnon@gmail.com>

Ramkumar Ramachandra <artagnon@gmail.com> writes:

> When --continue is invoked without any changes, the following stray
> error message appears- sed: can't read $dotest/final-commit: No such
> file or directory. Remove this by making sure that the file actually
> exists.

Show 16 quoted lines

>
> Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
> ---
>  git-am.sh |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/git-am.sh b/git-am.sh
> index 04f02a8..e61f47a 100755
> --- a/git-am.sh
> +++ b/git-am.sh
> @@ -693,7 +693,7 @@ do
>  	else
>  	    action=yes
>  	fi
> -	FIRSTLINE=$(sed 1q "$dotest/final-commit")
> +	test -e "$dotest/final-commit" && FIRSTLINE=$(sed 1q "$dotest/final-commit")

This will let the command follow the same codepath as before but the change in behaviour is that it does not reset FIRSTLINE to an empty string. Does this difference affect what the user sees after this part of the code?

I would generally prefer to use "test -f" not "-e" (simply because it came later and I am an old-timer), but if you want to use something newer than "test -f", it might be worth using "test -r" instead, as readability (not existence) is what you are after anyway.

Show 5 quoted lines

>  
>  	if test $action = skip
>  	then
> -- 
> 1.7.1