So, how does this work for non / files? I didn't think the lofi devices would work read/write either for somewhat similar reasons. Is there some sort of copy on write that happens there?
-jan Jan Setje-Eilers wrote: > Jack Schwartz wrote: >> Hi Jan SE. >> >> Thanks for confirming that you didn't see the update limitation as a bug. >> >> We'll go with that for now. ... and I'll just ask the question: if we >> filed a bug to get dcfs update working, what are the chances this >> enhancement could be made? > > Low. Figuring out how to make zfs a viable solution for small > compressed filesystems seems like a much more likely approach. > > -jan > > >> Thanks, >> Jack >> >> On 12/18/08 14:25, Jan Setje-Eilers wrote: >>> jan damborsky wrote: >>>> Hi Jack, >>>> >>>> >>>> Jack Schwartz wrote: >>>>> When I spoke yesterday to Jan Setje-eilers I wasn't under the >>>>> impression that he thought this was a dcfs bug to be fixed. Being >>>>> that each file is compressed individually, he said just don't >>>>> compress the files which will be needed for update. >>>> >>>> To be honest, I am not sure if this is the good long term approach >>>> to deal with this dcfs(7F) limitation. >>>> >>>> As we found out hitting this problem can generate issues which >>>> are not quite obvious they are caused by dcfs(7F) and thus might >>>> be not quite straightforward to evaluate/debug. >>>> >>>> Apparently, we can't always determine in advance which files need >>>> to be put on the list until we hit the problem which is manifestation >>>> of not having particular file on the list containing entries which >>>> are not compressed. >>>> >>>> I agree with you that this can be considered rather enhancement >>>> giving the fact that dcfs(7F) was initially considered to be used >>>> in read-only mode (for legacy miniroot), but since microroot is >>>> writable, my feeling is it might be better solve that problem on >>>> dcfs(7F) >>>> level once, rather than maintain special list of files forever. >>> >>> This is not a bug in dcfs. dcfs fundamentally doesn't work that way. >>> You're asking for seamlessly compressing ufs. I don't know of any >>> plans to add such support to ufs (or any other feature to ufs) for >>> that matter. >>> >>> -jan
