Date: Thu, 30 Aug 2012 16:36:56 -0400 (EDT) From: Rick Macklem <rmacklem@uoguelph.ca> To: Tim Gustafson <tjg@soe.ucsc.edu> Cc: freebsd-fs@freebsd.org Subject: Re: Using AMD with NFS Mounts Message-ID: <380236740.1347711.1346359016755.JavaMail.root@erie.cs.uoguelph.ca> In-Reply-To: <CAG27QgQOt%2BFD_Ef5HuiJV2euTzPLXk-nNN8iJ1L-2DmU0fKPxg@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Tim Gustafson wrote: > > Although I'd see it as an option initially, planning on supporting > > both > > "forever" would just result in bloat and yet more software to try > > and > > maintain, imho. (Theoretical comment, at least until a solid autofs > > port > > exists.) > > I don't see that as a valid argument against adding an autofs port. > Should we drop the port for Exim or Postgres, because ports for > Sendmail and mySQL exist? > The problem with something that goes in the kernel and uses the VFS/VOP interface is that it easily "bit rots" if it isn't in src/sys and being maintained by someone. (Note the file systems that aren't MP safe or don't work well that currently live within src/sys. Some will go away soon if the aren't made MP safe by someone.) I think you'd find that an autofs port (not necessarily in "ports") would suffer the same fate. (I'm honestly surprised that amd keeps working, since no one really maintains it. I mentioned that it is the last use of the old mount(2) syscall for NFS mounts. I know because I broke the backwards compatibility for mount(2) and amd got broken by that;-) However, I agree it isn't a reason for not having a port of it. I'd just rather see it replace amd, because it is old code I don't know how to maintain. This is similar to mountd.c. I am about to look at a couple of patches for mountd.c, but I hate doing this, because is it really old hard to read code. Lots of people complain about the lack of atomic export updates, but replacing mountd with nfse doesn't seem to be in the cards because it isn't absolutely 100% (including the bugs) compatible with mountd. (It seems mountd will exist "forever" unless someone implements a 100% compatible replacement and it will only get harder to maintain. If you don't believe me, take a look at mountd.c.:-) rick > -- > > Tim Gustafson > tjg@soe.ucsc.edu > 831-459-5354 > Baskin Engineering, Room 313A
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?380236740.1347711.1346359016755.JavaMail.root>