> He has a spurious dir_index filesystem optio, which seems to be new in > e2fsprogs 1.30. Anybody knows what that is, if we should implement this, > or if we better write a bug report ASAP to disable this for -o hurd?
It is listed in the feature_compat flags word, which is supposed to indicate that it's compatible with a filesystem that doesn't know about it. As far as I can tell, the Linux ext2fs patches that use this flag have not been integrated into Linux. So many Linux users are living happily without the functionality. With a little searching I've found an old version of the patch. It stores a hash table in directories alongside the normal linear records, so lookups can be done by a hash table lookup instead of the normal linear search. At least the reading of this format should be quite easy to implement in the Hurd's ext2fs format. The issues with writing the new flavor of directory blocks robustly could be more complex, which might have something to do with why this code has never made it into the Linux kernel mainline. _______________________________________________ Bug-hurd mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-hurd
