From: Spencer E. Olson <olsonse@umich.edu>
Date: Sun, 08 Aug 2010 03:57:35 GMT
Subject: Re: [PATCH v2] Allow HTTP user agent string to be modified.
Message-ID: <1281239855.7520.3.camel@compaq.olson.home>
In-Reply-To: <AANLkTi=G=GvxXEh1hGuYQK42cdkMUaFhqxsQ39K=2180@mail.gmail.com>

On Sun, 2010-08-08 at 10:51 +0800, Tay Ray Chuan wrote:

> >  static char *user_name, *user_pass;
> > +static const char *user_agent = NULL;
> 
> This can be skipped, I think.

My only reason for wanting to initialize is because I don't trust all
compilers to follow the standard.  But, since there are so many other
instances of relying on the compiler to zero out static vars in other
places already, I don't have any problem with conforming...


> 
> >[snip]
> > @@ -279,7 +283,10 @@ static CURL *get_curl_handle(void)
> >        if (getenv("GIT_CURL_VERBOSE"))
> >                curl_easy_setopt(result, CURLOPT_VERBOSE, 1);
> >
> > -       curl_easy_setopt(result, CURLOPT_USERAGENT, GIT_USER_AGENT);
> > +       if (user_agent != NULL)
> > +               curl_easy_setopt(result, CURLOPT_USERAGENT, user_agent);
> > +       else
> > +               curl_easy_setopt(result, CURLOPT_USERAGENT, GIT_USER_AGENT);
> 
> Hmm, perhaps
> 
>     curl_easy_setopt(result, CURLOPT_USERAGENT,
>         user_agent ? user_agent : GIT_USER_AGENT);
> 
> to replace the if-else?

That is perfectly fine with me.