On Mon, Jan 20, 2014 at 11:42:54PM +0100, Arkadiusz MiÅkiewicz wrote:
> On Monday 20 of January 2014, Dave Chinner wrote:
> > On Mon, Jan 20, 2014 at 06:22:48PM +0100, Arkadiusz MiÅkiewicz wrote:
> > > Hello.
> > >
> > > What's needed for both 3.1.12 and 3.2.0 releases to happen?
> > >
> > > If 3.2.0 is to be released soon then nice. Dave, AFAIK it is waiting for
> > > you and xfs_db crc work only?
> > xfs_repair work, actually. xfs_db is pretty much complete.
> > > Obviously I hope 3.2.0 is soon and there is no need to waste effort on
> > > 3.1.x.
> > The problem for me right now is that there are only so many hours in
> > the day....
> Could 3.2.0 be released without full xfs_db crc support then? And defer
> to 3.2.1 or even later. Especially you say "xfs_db is pretty much complete".
* release without full /xfs_repair/ support, not xfs_db.
No, I don't think so. releasing without full xfs_db support isn't an
issue as it's not critical to users. OTOH, xfs_repair is critical to
> Right now we have a situation where some features exist in kernel (as stable
> feature - separate project quota) that are not supported by userspace.
Separate project quota is v5 superblocks only, which are still
marked EXPERIMENTAL. For everything non-v5 superblock, v3.1.11 works
fine (modulo bugs).
> fixes (growfs thing/superblock garbage) missing etc. Improvements
> (performance/paralellism) waiting.
Nobody is going to push new features back to 3.1.x - we don't have
neough developer resources available to do that sort of work. That's
pretty obvious, given we are even having this conversation.
> ps. I assume that someone else does the actual release, so no need for more
> hours/day for you in such case.
Well, that's always been the plan since a 3.1.12 release was
proposed 3 months ago. How well has that plan been working out so