Re: Please pull the patch series "use the $( ... ) construct for command substitution"

Eric Sunshine [off-list ref] writes:

On Wed, May 14, 2014 at 1:14 PM, Matthieu Moy

quoted

I do not understand the use of the \ in front of the ` in the original
code.

The second argument of test_expect_success is double-quoted, so a bare
`...` would be evaluated before test_expect_success is even invoked.
Quoting it as \'...\' correctly suppresses the automatic evaluation,
giving test_expect_success the opportunity to evaluate it on-demand.

Ah, of course, you're right.

quoted

The correct code should be

        test x\"$(sed -n -e 4p < file)\" = x4 &&

I guess.

Same issue. The $(...) is being evaluated even before
test_expect_success is invoked. Better would be to suspend evaluation
via \$(...) to allow test_expect_success to evaluate it on-demand.

OK, then the correct code should be

        test x\"\$(sed -n -e 4p < file)\" = x4 &&

And the other attempt was close:

quoted

 test_expect_success 'verify pre-merge ancestry' "
-       test x\`git rev-parse --verify refs/heads/svn^2\` = \
-            x\`git rev-parse --verify refs/heads/merge\` &&
+       test x\$(git rev-parse --verify refs/heads/svn^2\) = \
+            x\$(git rev-parse --verify refs/heads/merge\) &&

\-escaping the $ was right, but the \) should be a single ).

In any case, the fact that we need to discuss this supports my claim
"this should be reviewed as a separate series".

-- 
Matthieu Moy
http://www-verimag.imag.fr/~moy/