On Thu, Aug 22, 2013 at 5:59 PM, Eric Sunshine <sunshine@xxxxxxxxxxxxxx> wrote: > On Thu, Aug 22, 2013 at 5:43 PM, Junio C Hamano <gitster@xxxxxxxxx> wrote: >> Eric Sunshine <sunshine@xxxxxxxxxxxxxx> writes: >>> On Thu, Aug 22, 2013 at 5:32 PM, Junio C Hamano <gitster@xxxxxxxxx> wrote: >>>> Now, I am curious how it breaks on OS X. >>>> >>>> My suspition is that "ignore_case" may have something to do with it, >>>> but what 2eac2a4c (ls-files -k: a directory only can be killed if >>>> the index has a non-directory, 2013-08-15) uses are the bog-standard >>>> cache_name_exists() and directory_exists_in_index(), so one of these >>>> internal API implementation has trouble on case insensitive >>>> filesystems, perhaps? I dunno. >>> >>> That's exactly my suspicion at the moment. It's an obvious difference >>> between Linux and OS X. I'm just in the process of trying to compare >>> between the two platforms. >> >> Or perhaps de->d_type does not exist? In such a case, we end up >> doing get_index_dtype() via get_dtype(), but in this codepath I >> suspect that we do not want to. We are interested in the type of >> the entity on the filesystem. > > de->d_type exists on both platforms. get_dtype() is never called. > > However, I did discover that treat_path() is being invoked fewer times > on OSX than on Linux. For instance, in the repository created by > t3010, treat_path() is called 19 times on Linux, but only 17 times on > OSX. Status update: For the 'pathx' directory created by the t3010 test, directory_exists_in_index() returns false on OSX, but true is returned on Linux. -- 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