From: Jeff King <peff@peff.net>
Date: Wed, 16 Mar 2016 17:31:28 GMT
Subject: Re: [PATCH] help.c: strip suffix only if the STRIP_EXTENSION defined
Message-ID: <20160316173127.GD4039@sigill.intra.peff.net>
In-Reply-To: <1458138449-26690-1-git-send-email-kuleshovmail@gmail.com>

On Wed, Mar 16, 2016 at 08:27:29PM +0600, Alexander Kuleshov wrote:

> We stripping extension in the list_commands_in_dir() to get
> commands without '.exe' suffix. Let's do it only if STRIP_EXTENSION
> is defined to not spend time for unnecessary strip_suffix() call in
> this case.
> 
> Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
> ---
>  help.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/help.c b/help.c
> index 19328ea..c865991 100644
> --- a/help.c
> +++ b/help.c
> @@ -153,8 +153,9 @@ static void list_commands_in_dir(struct cmdnames *cmds,
>  			continue;
>  
>  		entlen = strlen(ent);
> -		strip_suffix(ent, ".exe", &entlen);
> -
> +#ifdef STRIP_EXTENSION
> +		strip_suffix(ent, STRIP_EXTENSION, &entlen);
> +#endif

This is billed as an optimization in the commit message, but these two
pieces of code are not the same. The original always strips ".exe",
whether or not STRIP_EXTENSION is defined, and whether or not it is
".exe".

In practice it works out because people on Unix systems do not have
"git-foo.exe", and nobody sets STRIP_EXTENSION to other things.  But I
tend to think this is an improvement in robustness.

I also wonder if this should be sharing the strip_extension() helper
added in your 63ca1c0.

-Peff