Re: [PATCH 5/7] Fix tests under GETTEXT_POISON on pack-object

Jonathan Nieder [off-list ref] writes:

Nguyen Thai Ngoc Duy wrote:

quoted

On Tue, Aug 21, 2012 at 12:17 PM, Jonathan Nieder [off-list ref] wrote:

quoted

quoted

quoted

@@ -35,7 +35,7 @@ test_expect_success 'upload-pack fails due to error in pack-objects packing' '
      printf "0032want %s\n00000009done\n0000" \
              $(git rev-parse HEAD) >input &&
      test_must_fail git upload-pack . <input >/dev/null 2>output.err &&
-     grep "unable to read" output.err &&
+     test_i18ngrep "unable to read" output.err &&
      grep "pack-objects died" output.err

Wouldn't it make sense to change the second "grep" of output intended
for humans to test_i18ngrep while at it?

This comes from error("git upload-pack: git-pack-objects died with
error.") in unpack-trees.c, which is not i18n-ized yet. There's
another test in t5530 that does the same grep. I think we should leave
it as is until we mark the string for translation, then gettext poison
will spot it (verified) and we can fix it.

I don't understand the distinction you're making.  Isn't the message
intended for humans, and wouldn't changing that one line to
test_i18ngrep now save trouble later?  Tests are meant to check git's
intended behavior, not to exactly match its current behavior.

Correct; your suggestion is good.