Re: [PATCH v5 15/15] fast-export: don't handle uninteresting refs

From: Felipe Contreras <felipe.contreras@gmail.com>
Date: Nov 21, 2012, 07:11 UTC
Message-ID: <CAMP44s3vO8q+EW2sUUo6tLCQvD0PB4v_hZ9ySZjMD7wG2M8iwQ@mail.gmail.com>
In-Reply-To: <7vfw43pmp7.fsf@alter.siamese.dyndns.org>

On Wed, Nov 21, 2012 at 6:08 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 10 quoted lines

> I see Felipe keeps repeating that there are bugs, and keeps posting
> patches to change fast-export, but I haven't seen a concrete "No,
> the reason why you see these problems is because you are not using
> the interface correctly; the currrent interface is fine.  Here is
> how you can fix your program" from "others".
>
> With such a one-sided discussion, I've been having a hard time
> convincing myself if Felipe's effort is making the interface better,
> or just breaking it even more for existing remote helpers, only to
> fit his world model better.

IIRC you mentioned something about this mailing list being focused on *technical* merit. I've explained as much as I could, but at the end of the talk, talk is cheap, the code speaks for itself. I added a new very very very simple testgit remote helper, so anybody can see what's going on, and figure out how the interface could be used wrong.

Anybody can modify the bash version of git-remote-testgit and say 'no, the interface is not broken, here is how you push and pull without marks'. How hard is it to hack 82 lines of bash code?

But lets assume my testgit is fatally broken, would you, Junio, accept these patches if I show the same broken behavior with the python git-remote-testgit?

I'm afraid I have to point out the hard truth; the reason why nobody is doing that is because a) the interface is truly broken b) if they try, they most likely would fail, and that would prove they were wrong in previous discussion, or c) not enough familiarity with the code. I don't want to point fingers, nor do I intend to offend anybody, but I cannot find any other explanation of why this patch series, which is obviously correct (to me), doesn't receive any feedback, even though in theory, it should be very very very easy to show what's wrong with the series.

The tests are there, and the remote helper is as simple as it gets. There's nothing else but fast-export and transport-helper to blame for the issues. It's that simple.

Cheers.

-- 
Felipe Contreras