One of this developer's primary tools to diagnose broken regression tests is to run the test script using 'sh -x t... -i -v' to find out *which* call *actually* demonstrates the symptom. Hence it is pretty counterproductive if the test script behaves differently when being run via 'sh -x', in particular when using test_cmp or test_i18ncmp on redirected stderr. So let's use test_i18ngrep (as suggested by Jonathan Nieder) instead of test_cmp/test_i18ncmp to verify that stderr looks as expected. Signed-off-by: Johannes Schindelin <johannes.schindelin@xxxxxx> --- t/t1300-repo-config.sh | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) diff --git a/t/t1300-repo-config.sh b/t/t1300-repo-config.sh index 8867ce1..dca27a3 100755 --- a/t/t1300-repo-config.sh +++ b/t/t1300-repo-config.sh @@ -699,17 +699,13 @@ test_expect_success 'invalid unit' ' echo 1auto >expect && git config aninvalid.unit >actual && test_cmp expect actual && - cat >expect <<-\EOF && - fatal: bad numeric config value '\''1auto'\'' for '\''aninvalid.unit'\'' in file .git/config: invalid unit - EOF test_must_fail git config --int --get aninvalid.unit 2>actual && - test_i18ncmp expect actual + test_i18ngrep "bad numeric config value .1auto. for .aninvalid.unit. in file .git/config: invalid unit" actual ' test_expect_success 'invalid stdin config' ' - echo "fatal: bad config line 1 in standard input " >expect && echo "[broken" | test_must_fail git config --list --file - >output 2>&1 && - test_cmp expect output + test_i18ngrep "bad config line 1 in standard input" output ' cat > expect << EOF -- 2.7.4.windows.1 -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html