Re: [PATCH] http-backend: give a hint that web browser access is not supported

From: Jan Engelhardt <jengelh@inai.de>
Date: Dec 5, 2021, 10:13 UTC
Message-ID: <34pqs81o-36p6-s416-s791-1onsqo1734oo@vanv.qr>
In-Reply-To: <xmqqtufnonor.fsf@gitster.g>

On Sunday 2021-12-05 02:17, Junio C Hamano wrote:

Show 6 quoted lines

>>>What is in "pathinfo" parameter?
>> It is getenv("PATH_INFO").
>
>That part I know.  The question was what would a typical value of
>that parameter look like in the context of somebody mistakenly
>visiting Git smart HTTP endpoint via their browser.

As far as I can tell, it contains the request URI plus index.html resolution; https://git.inai.de/ reports /index.html while https://git.inai.de/foo reports /foo (since foo does not exist in the fs).

>I am basically wondering if it is helping the user enough, or if it
>is sufficient to give just the "err" and "hint", and nothing else.

I felt that, because ls(1) reports the filename again, e.g.

$ ls x ls: cannot access 'x': No such file or directory

that git-http-backend could do the same, especially since pathinfo isn't just $ENV{REQUEST_URI} again at all times.

Show 9 quoted lines

>> Yes, that seems more like it. I was not aware of send_strbuf.
>
>Heh, I wasn't either.  The review of this topic was the first time I
>seriously read any part of that file, and I think I still only read
>just about 20% of it ;-)
>
>Also, will the real Git clients, which are the primary intended
>audiences this program is trying to talk to, be OK if we suddenly
>start giving a non-empty 404 page?

I am confident enough to say yes. It's not like git-http-backend returned anything previously in the 404 case (like JSON or so), therefore clients could not possibly depend on content.

>If any implementations of Git HTTP client this program is serving
>(1) uses a 404 response as a cue to decide its next request
>(e.g. there may be some "try this URL and if it fails, do another
>one" fallback logic)

Not sure if they heed Location: headers, but I am not changing that :-)