git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: "Ryan Hendrickson via GitGitGadget" <gitgitgadget@gmail.com>
Cc: git@vger.kernel.org,  Ryan Hendrickson <ryan.hendrickson@alum.mit.edu>
Subject: Re: [PATCH] http: do not ignore proxy path
Date: Fri, 26 Jul 2024 09:29:54 -0700	[thread overview]
Message-ID: <xmqqr0bgw1z1.fsf@gitster.g> (raw)
In-Reply-To: <pull.1767.git.1722009170590.gitgitgadget@gmail.com> (Ryan Hendrickson via GitGitGadget's message of "Fri, 26 Jul 2024 15:52:50 +0000")

"Ryan Hendrickson via GitGitGadget" <gitgitgadget@gmail.com> writes:

> From: Ryan Hendrickson <ryan.hendrickson@alum.mit.edu>
>
> The documentation for `http.proxy` describes that option, and the
> environment variables it overrides, as supporting "the syntax understood
> by curl". curl allows SOCKS proxies to use a path to a Unix domain
> socket, like `socks5h://localhost/path/to/socket.sock`. Git should
> therefore include, if present, the path part of the proxy URL in what it
> passes to libcurl.
>
> Signed-off-by: Ryan Hendrickson <ryan.hendrickson@alum.mit.edu>
> ---
>     http: do not ignore proxy path
>     
>      * Documentation: do I need to add anything?

http.proxy documentation says

  The syntax thus is [protocol://][user[:password]@]proxyhost[:port].

but the updated code pays attention to what can come after the
"host[:post]" part, does it not?

> diff --git a/http.c b/http.c
> index 623ed234891..0cd75986a6b 100644
> --- a/http.c
> +++ b/http.c

I am unfamiliar with this code path, so let me follow along aloud.

> @@ -1265,7 +1265,13 @@ static CURL *get_curl_handle(void)
>  		if (!proxy_auth.host)
>  			die("Invalid proxy URL '%s'", curl_http_proxy);

We grabbed the value from the configuration variable (or various
environment variables like $http_proxy) in the curl_http_proxy
variable, and then passed it to credential_from_url() to parse parts
out of [protocol://][user[:password]@]proxyhost[:port][/p/a/t/h].
The parsed result is in proxy_auth struct and there is no hope it
can be usable if the .host member is missing.

> -		curl_easy_setopt(result, CURLOPT_PROXY, proxy_auth.host);

We used to only use the .host member but the curl_http_proxy could
have had a path after it, held in the .path member.

> +		if (proxy_auth.path) {
> +			struct strbuf proxy = STRBUF_INIT;
> +			strbuf_addf(&proxy, "%s/%s", proxy_auth.host, proxy_auth.path);
> +			curl_easy_setopt(result, CURLOPT_PROXY, proxy.buf);
> +			strbuf_release(&proxy);
> +		} else
> +			curl_easy_setopt(result, CURLOPT_PROXY, proxy_auth.host);

Style.  If "if" side needs {braces} because it consists of multiple
statements, the corresponding "else" side should also have {braces}
around its body, even if it only has a single statement.

If you have the proxy strbuf in a bit wider scope, then the above becomes

	if (proxy_auth.path)
		strbuf_addf(&proxy, "%s/%s", proxy_auth.host, proxy_auth.path);
	else
		strbuf_addstr(&proxy, proxy_auth.host);
	curl_easy_setopt(result, CURLOPT_PROXY, proxy.buf);
	strbuf_release(&proxy);

which might (I am not decided) be easier to follow.

Could existing users have been taking advantage of the fact that the
extra /path at the end of http.proxy (and $http_proxy and friends)
are ignored?  For them, this change will appear as a regression.

Other than that (and the lack of any documentation and test
updates), looking good.

Thanks.

  reply	other threads:[~2024-07-26 16:29 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-26 15:52 [PATCH] http: do not ignore proxy path Ryan Hendrickson via GitGitGadget
2024-07-26 16:29 ` Junio C Hamano [this message]
2024-07-26 17:12   ` Ryan Hendrickson
2024-07-26 17:45     ` Junio C Hamano
2024-07-26 21:11 ` Jeff King
2024-07-26 22:43   ` Ryan Hendrickson
2024-07-29 19:31     ` Jeff King
2024-07-27  6:44 ` [PATCH v2] " Ryan Hendrickson via GitGitGadget
2024-07-29 20:09   ` Jeff King
2024-07-31 15:33     ` Ryan Hendrickson
2024-07-31 16:01   ` [PATCH v3] " Ryan Hendrickson via GitGitGadget
2024-07-31 22:24     ` Junio C Hamano
2024-08-01  3:44       ` Ryan Hendrickson
2024-08-01  5:21         ` Junio C Hamano
2024-08-01  5:45       ` Jeff King
2024-08-01 14:40         ` Junio C Hamano
2024-08-01  5:22     ` [PATCH v4] " Ryan Hendrickson via GitGitGadget
2024-08-01  6:04       ` Jeff King
2024-08-01 17:04         ` Junio C Hamano
2024-08-02  5:20       ` [PATCH v5] " Ryan Hendrickson via GitGitGadget
2024-08-02 15:52         ` Junio C Hamano
2024-08-02 16:43           ` Ryan Hendrickson
2024-08-02 17:10             ` Junio C Hamano
2024-08-02 18:03               ` Ryan Hendrickson
2024-08-02 19:28                 ` Junio C Hamano
2024-08-02 19:39                   ` Ryan Hendrickson
2024-08-02 21:13                     ` Junio C Hamano
2024-08-02 21:26                       ` Ryan Hendrickson
2024-08-02 21:43                         ` Junio C Hamano
2024-08-02 21:47                         ` Junio C Hamano
2024-08-02 22:14                           ` Ryan Hendrickson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=xmqqr0bgw1z1.fsf@gitster.g \
    --to=gitster@pobox.com \
    --cc=git@vger.kernel.org \
    --cc=gitgitgadget@gmail.com \
    --cc=ryan.hendrickson@alum.mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).