Am 24.03.2014 22:14, schrieb Jeff King:
> On Mon, Mar 24, 2014 at 11:22:58AM -0700, Junio C Hamano wrote:
>
>> René Scharfe <l.s.r@xxxxxx> writes:
>>
>>> -test_expect_success 'log --grep -i' '
>>> - git log -i --grep=InItial --format=%H >actual &&
>>> - test_cmp expect_initial actual
>>> -'
>>> +test_log expect_initial --grep initial
>>> +test_log expect_nomatch --grep InItial
>>
>> This, and the next --author one, assumes that we will never break
>> "--grep=foo" without breaking "--grep foo". That should be OK, but
>> we might want to add separate tests e.g.
>>
>> test_log expect_initial --grep=initial
>>
>> perhaps? I dunno.
>
> Yeah, I I'd prefer "--grep=" here (and in all scripts). In general, I
> think our attitude should be that "--foo=bar" is guaranteed to work
> forever, but "--foo bar" is not. The latter only works if the argument
> is non-optional, so that leaves us room to "break" compatibility to make
> an argument optional in a future version.
>
> Now, whether the rest of the world and its script-writers are aware of
> this fact, I don't know. So it may be too late already (but it does look
> like we mention it in gitcli(7)).
OK, then the following should be squashed into patch 2 (t4209: factor out
helper function test_log()):
diff --git a/t/t4209-log-pickaxe.sh b/t/t4209-log-pickaxe.sh
index 9f3bb40..f47231a 100755
--- a/t/t4209-log-pickaxe.sh
+++ b/t/t4209-log-pickaxe.sh
@@ -10,6 +10,14 @@ test_log() {
shift 3
rest=$@
+ case $kind in
+ --*)
+ opt=$kind=$needle
+ ;;
+ *)
+ opt=$kind$needle
+ ;;
+ esac
case $expect in
expect_nomatch)
match=nomatch
@@ -20,7 +28,7 @@ test_log() {
esac
test_expect_success "log $kind${rest:+ $rest} ($match)" "
- git log $rest $kind $needle --format=%H >actual &&
+ git log $rest $opt --format=%H >actual &&
test_cmp $expect actual
"
}
--
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