On Sun, May 20, 2012 at 9:10 AM, Jon Seymour <jon.seymour@xxxxxxxxx> wrote:
> On Sun, May 20, 2012 at 8:51 AM, Jon Seymour <jon.seymour@xxxxxxxxx> wrote:
>> On Sun, May 20, 2012 at 4:56 AM, Jens Lehmann <Jens.Lehmann@xxxxxx> wrote:
>>> Am 19.05.2012 06:40, schrieb Jon Seymour:
>> I noticed one relative case that is not handled properly yet, but
>> there is a workaround. If the superproject's origin URL is of the
>> form: foo/bar (a case I actually have myself for reasons I can explain
>> if you want me to), then the correct rule doesn't get matched by
>> .*/*). The workaround is for the user to change foo/bar style origin
>> URLs to ./foo/bar.
>>
>> Let me know if I should fix this case now too.
>
> I think this expression:
>
> remoteurl=$(echo "remoteurl" | sed "s|^[^/][^:]*\$|./&|" )
> remoteurl=${remoteurl%/}
>
> would normalize remoteurl correctly for the foo/bar case and not any other.
>
> I'd also need to add at least one new test (or do I actually need 3?),
> of course.
Ok. I have a version that implements this fix:
url="$1"
+ remoteurl=$(echo "$remoteurl" | sed "s|^[^/][^:]*\$|./&|")
remoteurl=${remoteurl%/}
and...
+ .*/*)
+ up_path="$(echo "$2" | sed "s/[^/]*/../g")"
+ remoteurl=${remoteurl#./}
+ remoteurl="${up_path%/}/${remoteurl%/*}"
+ ;;
with a single test that demonstrates that it works. The additional #./
tweak is so that submodule URLs end up like:
../foo/sub
instead of:
../foo/./sub
jon.
--
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