[02:46:55] --- Russ has become available
[02:51:08] <jhutz@jis.mit.edu/owl> Is the full spec there, or just the bits related to compressed dumps?
[03:13:16] --- Russ has left: Disconnected
[04:21:53] --- cclausen has left
[06:41:25] --- dev-zero@jabber.org has become available
[06:41:29] --- dev-zero@jabber.org has left: offline
[07:28:52] --- edgester has become available
[08:38:52] --- cclausen has become available
[10:00:27] --- haba has become available
[10:57:32] --- dev-zero@jabber.org has become available
[11:02:27] --- dev-zero@jabber.org has left: offline
[11:04:51] --- haba has left
[11:33:39] --- mdionne has become available
[11:43:43] <mdionne> if (Pos + Len > tlen) Len = tlen - Pos;       /* get length we should send */
Is this intentional that we return a negative length when fetching past the current end of file (in fileserver)?  The client
just seems to skip the read if the value is <= 0, so 0 makes it happy.  The negative length makes the client unhappy
in the CForeign case. 
[12:15:40] --- deason has become available
[12:22:44] <mdionne> With that fixed and after adding AFS_DFSSymlink(), a standard client is fully functional with per-file ACLs
[13:28:16] --- Russ has become available
[13:55:59] --- dev-zero@jabber.org has become available
[13:56:07] --- dev-zero@jabber.org has left: offline
[14:06:07] --- Russ has left: Replaced by new connection
[14:06:08] --- Russ has become available
[14:32:00] --- Russ has left: Disconnected
[16:12:24] --- cclausen has left
[17:11:16] --- RedBear has left
[18:00:30] <jhutz@jis.mit.edu/owl> > Is this intentional that we return a negative length when fetching past
> the current end of file (in fileserver)?
Two out of three people at this table agree, "yes but we don't remember
why"
[18:00:45] * edgester is reading the slides from the workshop. Harmut does some crazy cool stuff
[18:06:01] <jhutz@jis.mit.edu/owl> Yes; he's scary
[18:06:45] <edgester> It's sad that he has to do these tricks and that openafs doesn't have the performance
[18:06:58] <edgester> that he needs
[18:09:10] <jhutz@jis.mit.edu/owl> No; going through an intermediary will never have the performance or
scalability of being able to talk directly to the disks.
[18:09:27] <edgester> bummer
[18:10:24] <edgester> ok, but could we get to within 10%-20% of that native disk perfomance?
[18:27:27] --- Rrrrred has left
[19:03:41] --- RedBear has become available
[19:09:42] --- mdionne has left
[19:11:06] --- edgester has left
[19:44:31] --- cclausen has become available
[23:25:18] --- deason has left