From: Junio C Hamano <gitster@pobox.com>
Date: Tue, 16 Feb 2016 21:20:57 GMT
Subject: Re: [PATCH +warn] Implement https public key pinning
Message-ID: <xmqqr3gcmjwm.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <20160216032818.GB19954@sigill.intra.peff.net>

Jeff King <peff@peff.net> writes:

> On Mon, Feb 15, 2016 at 07:19:07PM -0800, Junio C Hamano wrote:
>
>> I suspect that "#else" is too agressive to bail out or something
>> silly like that.
>> 
>> Oh, I think I found it.
>> 
>> @@ -216,6 +219,13 @@ static int http_options(const char *var, const char *value, void *cb)
>>  	if (!strcmp("http.sslcapath", var))
>>  		return git_config_pathname(&ssl_capath, var, value);
>>  #endif
>> +	if (!strcmp("http.pinnedpubkey", var))
>> +#if LIBCURL_VERSION_NUM >= 0x072c00
>> +		return git_config_pathname(&ssl_pinnedkey, var, value);
>> +#else
>> +		warning(_("Public key pinning not supported with cURL < 7.44.0"));
>> +		return 0;
>> +#endif
>> 
>> We are not writing in Python.  Indenting the second line the same
>> way does not make it part of the block.  Of course by inserting the
>> new config in the earlier part of the function, it broke everything
>> that comes after.
>
> Oof. Good eyes. I suspected something funny with the #if, but after
> starting at it for minutes, couldn't see it.
>
> That makes sense, and the fix is obvious.

Yup.  I think I sent a patch already and I followed the "when adding
a new thing to a set, add it at the end unless there is other reason
to place it elsewhere".

Thanks.