[00:08:06] --- Russ has left: Disconnected [02:20:12] Who added the 1.6 and 1.4 branches to Ohloh? Because it's completely screwing up our statistics - commits and lines of code are triple counted, and we're now flagged as having a "declining year on year development effort". [04:22:14] not me [04:35:49] --- haba has left [05:10:36] --- haba has become available [06:00:21] --- steven.jenkins has become available [06:20:07] --- steven.jenkins has left [06:20:20] --- steven.jenkins has become available [06:34:47] Crikey, we use assert(X) in a lot of places where X really, really, does need to happen. [06:36:00] we should be using osi_Assert in those places. [06:36:13] that was... sort of the point of osi_Assert [06:37:31] I thought the point of osi_Assert was that the assertion expression went into the logs for that service, rather than just going to stderr and being lost. [06:38:06] well, my point was that osi_Assert doesn't honor NDEBUG. [06:38:07] But yeah. In fact, I'd argue that we should have osi_Assert() and osi_Verify(), one that has the same behaviour as assert() and the other that will always execute the expression. [06:38:25] I have a patch for the Unix code. [06:41:45] The one drawback with osi_Assert is that it creates a dependency on RX. But I have a fix for that, too. [06:58:30] --- meffie has become available [06:59:58] I think that you should implement opr_Assert and opr_Verify and make osi_Assert be a pass-through to opr_Assert and then deprecate it. [07:03:46] We need osi_Assert to have special meaning in the bits of code that might get compiled in kernel space - but yeah, my rough plan is that we have opr_Verify and opr_Assert for all non kernel code. And code that is mixed kernel/userspace, osi_Assert == opr_Assert when built in userspace. [07:07:36] And assert(1 == 2) becomes abort() [07:08:42] --- reuteras has left [07:24:30] --- deason/gmail has become available [07:28:29] --- Simon Wilkinson has left [08:29:30] Derrick, I don't suppose you looked at whether the getevent() implementations were merge candidates while reviewing shutdown_osisleep? I realized yesterday that I should have done so. [08:29:48] (Also, it would be great if gerrit could show me the diff between patchset N and patchset M) [08:36:56] it can show you individual files via e.g. unified diff -> patch history [08:37:16] the whole patchset diff can be viewed via git or gitweb [08:37:28] (I could just use git, but I don't have git on this machine.) [08:39:52] gitweb will if you.... (lemme see) a=commitdiff;hp=;h= [08:40:10] that's a diff of the trees of course, though, not like a diff of diffs or anything [09:32:55] i didn't, no. [09:33:22] Shrug. It could always be a separate change. [09:33:33] which would be fine [10:30:15] --- haba has left [10:54:53] --- Russ has become available [10:55:40] --- jaltman/FrogsLeap has left: Disconnected [10:55:44] --- steven.jenkins has left [10:59:03] --- steven.jenkins has become available [11:01:53] --- jaltman/FrogsLeap has become available [11:22:52] --- steven.jenkins has left [11:23:04] --- steven.jenkins has become available [14:13:03] --- meffie has left [16:01:14] --- deason/gmail has left [16:34:26] --- jaltman/FrogsLeap has left: Disconnected [16:43:56] --- jaltman/FrogsLeap has become available [17:00:28] --- pod has left [21:03:01] --- jaltman/FrogsLeap has left: Replaced by new connection [21:03:02] --- jaltman/FrogsLeap has become available [21:09:13] --- jaltman/FrogsLeap has left: Disconnected [21:09:34] --- jaltman/FrogsLeap has become available