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

On Wed, Jun 27, 2018 at 2:27 AM Johannes Sixt [off-list ref] wrote:

Am 27.06.2018 um 04:15 schrieb Elijah Newren:

quoted

On Tue, Jun 26, 2018 at 2:01 PM, Jeff King [off-list ref] wrote:

quoted

On Tue, Jun 26, 2018 at 04:46:18PM -0400, Eric Sunshine wrote:

quoted

Some of these dangers can be de-thoothed during the linting phase by
defining do-nothing shell functions:
     cp () { :; }
That, at least, makes the scariest case ("rm") much less so.

Now that's an interesting idea. We can't catch every dangerous action
(notably ">" would be hard to override), but it should be pretty cheap
to cover some obvious ones.

Crazy idea: maybe we could defang it a little more thoroughly with
something like the following (apologies in advance if gmail whitespace
damages this):

-               if test "OK-117" != "$(test_eval_ "(exit 117) &&
$1${LF}${LF}echo OK-\$?" 3>&1)"
+               if test "OK-117" != "$(test_eval_ "cd() { return 0; }
&& PATH=/dev/null && export PATH && (exit 117) && $1${LF}${LF}echo
OK-\$?" 3>&1)"

Interesting idea (coupled with Hannes's point below)...

I'd define all these functions as { return 1; } because we want to stop
any && chain as early as possible (and with an exit code that is not the
sentinel value).

A very sensible suggestion.