[01:01:56] --- squinney has become available
[04:45:30] --- Stephan Wiesand has become available
[04:55:39] --- dev-zero@jabber.org has become available
[05:04:57] --- mlane has become available
[05:05:50] --- mlane has left
[05:16:16] --- dev-zero@jabber.org has left
[05:49:49] --- Stephan Wiesand has left
[06:29:31] --- meffie has become available
[06:53:12] --- squinney has left
[07:03:44] --- ballbery has become available
[08:14:29] <kaduk@mit.edu/barnowl> clang -Wall -Wextra whines at me because rxi_GetCallNumberVector takes
an afs_int32* and we declare the call numbers to be unsigned in the
rxgk spec.  Do we ever use negative call numbers?
[08:55:12] <Simon Wilkinson> No
[08:56:12] <kaduk@mit.edu/barnowl> Is it worth the churn to change the signature of the rx functions?
[08:57:51] <Simon Wilkinson> rxkad uses an array of int32s
[08:58:17] <kaduk@mit.edu/barnowl> That would be part of the churn, too
[08:58:43] <Simon Wilkinson> There are potentially on-the-wire encoding issues there.
[08:59:13] <kaduk@mit.edu/barnowl> How so?
[08:59:53] <Simon Wilkinson> That structure is sent on the wire. We'd need to be very careful that there aren't behaviour changes with large callNumbers
[09:00:16] <Simon Wilkinson> (although, to get a callNumber > INT_MAX, you'd need to be trying _really_ hard, and would  probably have already wrapped other stuff in RX)
[09:01:00] <Simon Wilkinson> The callNumber in the channel (formerly in the connection) is a signed value.
[09:01:07] <Simon Wilkinson> So, it's all a bit of a mess internally.
[09:02:18] <kaduk@mit.edu/barnowl> As with so many things...
[09:03:20] <Simon Wilkinson> Our signed/unsigned story is such a mess. Russ's opinion, based on his work with INN, was that fixing it all was likely to introduce more bugs than it solves.
[09:03:36] <kaduk@mit.edu/barnowl> INN?
[09:04:16] <Simon Wilkinson> http://www.isc.org/software/inn
[09:05:01] <kaduk@mit.edu/barnowl> Hmm.  I suppose I have an easy way to silence these warnings handy.
[09:05:22] <Simon Wilkinson> I suspect just casting the pointer will make them go away.
[09:05:33] --- meffie has left
[09:05:40] <Simon Wilkinson> GIven the mess internally with RX, I think you can assume that you don't care about call numbers that are larger than INT_MAX
[09:14:11] --- stephan.wiesand has become available
[09:25:33] --- ballbery has left
[09:25:48] --- ballbery has become available
[09:26:42] --- ballbery has left
[09:27:00] --- ballbery has become available
[11:00:17] --- meffie has become available
[11:28:38] --- ballbery has left
[11:29:43] --- ballbery has become available
[11:33:38] --- stephan.wiesand has left
[14:00:54] --- ballbery has left
[15:52:10] --- mdionne has become available
[16:18:15] --- meffie has left
[16:57:23] --- jaltman/FrogsLeap has left: Disconnected
[17:01:51] --- jaltman/FrogsLeap has become available
[17:03:33] --- jaltman/FrogsLeap has left: Disconnected
[17:09:23] --- jaltman/FrogsLeap has become available
[18:10:44] --- mdionne has left
[20:33:06] --- jaltman/FrogsLeap has left: Disconnected
[20:36:16] --- jaltman/FrogsLeap has become available
[23:03:19] --- Simon Wilkinson has left