Re: [PATCH] http: fix an unused variable warning for 'curl_no_proxy'

From: Jeff King <peff@peff.net>
Date: Mar 14, 2018, 22:15 UTC
Message-ID: <20180314221544.GA20167@sigill.intra.peff.net>
In-Reply-To: <517c4210-c381-899e-b13a-00f8e4caba74@ramsayjones.plus.com>

On Wed, Mar 14, 2018 at 09:56:06PM +0000, Ramsay Jones wrote:

Show 14 quoted lines

> Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
> ---
> 
> Hi Junio,
> 
> I happened to be building git on an _old_ laptop earlier this evening
> and gcc complained, thus:
> 
>       CC http.o
>   http.c:77:20: warning: ‘curl_no_proxy’ defined but not used [-Wunused-variable]
>    static const char *curl_no_proxy;
>                       ^
> The version of libcurl installed was 0x070f04. So, while it was fresh in my
> mind, I applied and tested this patch.

Makes sense. This #if would go away under my "do not support antique curl versions" proposal. I haven't really pushed that forward since Tom Christensen's patches to actually make the thing build (and presumably since he is building on antique versions he can't turn on -Werror anyway, since IIRC it tends to have some false positives).

I agree with Jonathan that this explanation should be in the commit message. The patch itself looks OK, although:

Show 19 quoted lines

> diff --git a/http.c b/http.c
> index 8c11156ae..a5bd5d62c 100644
> --- a/http.c
> +++ b/http.c
> @@ -69,6 +69,9 @@ static const char *ssl_key;
>  #if LIBCURL_VERSION_NUM >= 0x070908
>  static const char *ssl_capath;
>  #endif
> +#if LIBCURL_VERSION_NUM >= 0x071304
> +static const char *curl_no_proxy;
> +#endif
>  #if LIBCURL_VERSION_NUM >= 0x072c00
>  static const char *ssl_pinnedkey;
>  #endif
> @@ -77,7 +80,6 @@ static long curl_low_speed_limit = -1;
>  static long curl_low_speed_time = -1;
>  static int curl_ftp_no_epsv;
>  static const char *curl_http_proxy;
> -static const char *curl_no_proxy;

I'm not sure whether our ordering of these variables actually means much, but arguably it makes sense to keep the proxy-related variables near each other, even if one of them has to be surrounded by an #if.

I guess you were going for ordering the #if's in increasing version order. I'm not sure the existing code follows that pattern very well.

-Peff