From: erik elfström <erik.elfstrom@gmail.com>
Date: Tue, 01 Sep 2015 06:27:59 GMT
Subject: Re: [PATCH] t7300: fix broken && chains
Message-ID: <CAMpP7NZ=K62XVkuuJHpDruBy7b2gECHrJ30L43LzdE54vTVB2A@mail.gmail.com>
In-Reply-To: <20150831185414.GB20555@sigill.intra.peff.net>

On Mon, Aug 31, 2015 at 8:54 PM, Jeff King <peff@peff.net> wrote:
> On Sun, Aug 30, 2015 at 11:18:09AM +0200, Erik Elfström wrote:
>
> Unfortunately CHAIN_LINT cannot reach inside a nested subshell. I cannot
> think of a way to make it do so, and besides, that is also the way to
> override the &&-chain precedence. So I think it is not really possible
> to get better coverage here. And such cases as these are not really very
> interesting (e.g., here we exclude only minor minor setup steps from the
> &&-chain).
>
> -Peff

I actually hacked together an ugly script to check for more broken
chains. It's not working very well and I haven't checked the output
thoroughly but I did see quite a few broken chains. Many are of the
(mostly harmless) form:

     (
        echo "100644 $o5 0    a"
        echo "100644 $o0 0    c"
        echo "160000 $c1 0    d"
    ) >expected &&

I'd estimate that there are hundreds of these (see t3030 for
examples). I'm not sure if you care about these? As you say they are
not really very interesting cases.

/Erik