Hi Ryusuke, On Mon, 2014-04-21 at 02:05 +0900, Ryusuke Konishi wrote: [snip] > > There are two formats - seconds and human-readable format. > > You can set preferable time format by command > > (for example, setting human-readable format): > > > > 'echo human-readable > /sys/fs/nilfs/features/time_format' > > I don't think you shouldn't globally switch the time format like this. > I think two or more sysfs files should be added per format, or > either one should be selected if avoiding complexity. > > Think the situation where userland programs read time stamp > information on the sysfs interface. They will malfunction if a user > incidentally changes the format through the time_format file. The > format of sysfs files should never depend on other changeable status. > OK. I agree. So, I choose to have two sysfs files (for example, last_seg_write_time for human-readable format and last_seg_write_time_secs for output in seconds). > The features directory of sysfs interface should have global features > which are independent to nilfs instance or version (e.g. nilfs3, etc) > since you chose "nilfs" for the fs directory name. The revision file > is self-contradictory in that sense. I think it should be placed in > each device directory. Otherwise, we should use "nilfs2" as for the > fs directory name. > In current implementation: (1) fs/nilfs/features/revision - show current supported revision by file system driver. (2) fs/nilfs/<device>/revision - show file system revision that it saved in superblock of the volume. As a result, it shows the revision of file system is created on a volume. I suppose that we have identical understanding. And current implementation provides information about revision in the proper way. If I misunderstand something, please, correct me. Thanks, Vyacheslav Dubeyko. -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
