Hi Olaf, On Fri, 2011-06-24 at 17:27 +0200, Olaf van der Spek wrote: > On Thu, Mar 17, 2011 at 1:55 PM, Olaf van der Spek <[email protected]> > wrote: > > Hi Stewart, > > > > Just watched your talk "Dropping ACID" and wanted to respond, might be > > interesting for others on the list too. > > open, write, fsync, close, rename is not the atomic equivalent of > > open, write, close. It is atomic but not equivalent as it has a big > > number of assumptions / restrictions (losing meta-data, assuming > > permissions, assuming same volume). So programmers don't getting it > > right is to be expected. > > I argued for O_ATOMIC, a way to do atomic file data replacements, but > > the FS devs prefer to not solve this problem. I guess they don't care > > about reliability too much. :p > > Any comments? ;)
See the replies from Ted Ts'o on the subject when you asked on the debian-devel list in January. It isn't that they prefer not to solve the problem, just that the problem is not solvable without a large bulk of code (and/or problems) and it would give very little gain over fsync. Kind Regards -- Andrew Hutchings - LinuxJedi - http://www.linuxjedi.co.uk/ _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
