> I removed the additional check from the IsValid function of the > superblock, because while reading the linux implementation, it still mounts it normally (even if it has > errors) but gives out a warning that the file system should be verified. What would be the correct way to handle such errors (ie. check required)? Also, linux checks > the revision number, and if it's unrecognized, it only mounts the file system in read only mode. I assume this is for forward compatibility. Should I handle it the same > way? I think mounting it read-only and outputting the error to the syslog is a good way to warn in these cases > Also, I am currently implementing initial journal support, and I > think I'll do the idea to apply the journal changes in memory when the file system is mounted as read-only. Is it okay if I publish the incomplete code online somewhere so people can view and comment before I do an actual patch submission? I think it is. You can also send the patch to the mailing list, stating that you know it's not ready.