Re: [PATCH 29/29] t/test-lib: teach --chain-lint to detect broken &&-chains in subshells

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 26, 2018, 19:15 UTC
Message-ID: <xmqqwouljr5e.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20180626073001.6555-30-sunshine@sunshineco.com>

Eric Sunshine <sunshine@sunshineco.com> writes:

Show 26 quoted lines

> The --chain-lint option detects broken &&-chains by forcing the test to
> exit early (as the very first step) with a sentinel value. If that
> sentinel is the test's overall exit code, then the &&-chain is intact;
> if not, then the chain is broken. Unfortunately, this detection does not
> extend to &&-chains within subshells even when the subshell itself is
> properly linked into the outer &&-chain.
>
> Address this shortcoming by eliminating the subshell during the
> "linting" phase and incorporating its body directly into the surrounding
> &&-chain. To keep this transformation cheap, no attempt is made at
> properly parsing shell code. Instead, the manipulations are purely
> textual. For example:
>
>     statement1 &&
>     (
>         statement2 &&
>         statement3
>     ) &&
>     statement4
>
> is transformed to:
>
>     statement1 &&
>         statement2 &&
>         statement3 &&
>     statement4

so, with --chain-lint, we would transform this

	mkdir -p a/b/c &&
	(
		cd a/b/c
		rm -fr ../../*
	) &&
	statement 4

into this sequence

	(exit $sentinel) &&
	mkdir -p a/b/c &&
		cd a/b/c
		rm -fr ../../* &&
	statement 4

and then rely on the non-zero exit to cancel all the remainder?

We didn't create nor cd to the t/trash$num/a/b/c thanks to the && chain, and end up running rm -fr ../../* from inside t/trash$num?

Hmmmmm....