Re: Other tidbits
Shirish Hemant Phatak <[email protected]> Fri, 18 Oct 2002 17:08:24 -0400
| Newsgroups | gmane.comp.file-systems.intermezzo.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi, The entire EA and POSIX ACL replication for Intermezzo needs to be redone. The fundamental assumption of the old model was that ACLs would always be manipulated through the EA interface, which is no longer true. In many ways the new model considertably simplifies the old code and will not require many of the old ugly hacks that I had to use. I believe the Intermezzo module does compile correctly on 2.5; I am not sure it works just yet. Peter and Robert would know. I have partially done EA/ACL patch for the 2.4 tree, however, since then the tree has moved away from the original codebase much to rapidly for me to track. In general, the patch in not very difficult to do, but will require 2-3 days of dedicated effort for coding and testing; especially since there are lots of tiny details and issues when correctly replicating the ACLs and getting them to work just right (the EAs are easy). -Shirish Trever L. Adams wrote: >I am just curious about the ACLs and EAs as 2.5 now has them and I >believe the API is getting pretty stable. An additional question is why >rely on http servers instead of making a more stream lined protocol? > >Trever Adams > >On Mon, 2002-01-28 at 18:36, Shirish H. Phatak wrote: > > >>>>* NTFS or POSIX style ACLs >>>> >>>> I know, Linux doesn't yet have these integrated into the main kernel. >>>>However, I believe it will be in 2.5. >>>> >>>> >>>> >>>> >>>> >>>And InterMezzo already replicates POSIX ACL's. Shirish is the expert here. >>> >>>Thanks for your interest. >>> >>> >>> >> For some reason large chunks of the extended attribute replication >>code (which also includes POSIX acls) have been commented out, so I >>will have to go in once more and see whats going on. I have had ACL >>replication working for quite some time at Tacit. Once the 2.5 api >>stabilizes, a forward port will be very easy, this code is largely >>independant from the rest of the intermezzo kernel code. >> >>-Shirish >> >> >> >>>- Peter - >>> >>> > > > ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf