[01:33:50] --- dev-zero@jabber.org has left [01:45:25] --- dev-zero@jabber.org has become available [01:45:57] --- dev-zero@jabber.org has left [01:45:59] --- dev-zero@jabber.org has become available [01:59:31] --- dev-zero@jabber.org has left [02:33:14] --- dev-zero@jabber.org has become available [06:25:06] --- meffie has become available [06:25:48] --- ballbery has become available [06:45:25] --- meffie has left [06:47:25] --- meffie has become available [06:52:41] --- mvitale has become available [07:19:17] --- deason has become available [07:19:30] --- Stephan Wiesand has become available [07:50:03] --- Ken Dreyer has left [09:13:50] --- Stephan Wiesand has left [09:27:14] --- dev-zero@jabber.org has left [09:38:55] --- mvitale has left [09:50:30] --- mvitale has become available [10:26:02] --- dev-zero@jabber.org has become available [11:13:03] --- mvitale has left [11:20:46] --- mvitale has become available [11:30:01] --- stephan.wiesand has become available [11:48:54] --- mdionne has left [12:25:52] --- stephan.wiesand has left [14:03:09] --- meffie has left [14:15:35] --- ballbery has left [14:31:43] --- mvitale has left [14:38:20] --- mvitale has become available [14:48:19] --- ballbery has become available [14:48:33] --- ballbery has left [15:17:17] --- dev-zero@jabber.org has left [15:29:33] I see that opr_Assert does not guarantee its argument will be evaluated; does osi_Assert do so? (I think there maybe was mail about this?) [15:30:55] osi_Assert currently does so. [15:31:14] It's also the only assertion routine you can use in kernel code. [15:32:14] In userspace only code, you should use opr_Assert, or opr_Verify, according to the behaviour you require. [15:32:20] Right, I found "Further rationalise our usage of assert()" which mentions this. [15:33:30] Longer term, I suspect we'll end up making the same differentiation in the kernel. [15:33:35] But neither that nor the commit introducing opr_Assert mentions a guarantee, so I wanted to check. [15:34:34] --- deason has left [15:49:05] osi_Assert predates all of those changes. It goes back to the original IBM code. [15:49:25] Right. [15:50:00] I would not have been surprised if the commit introducing the new API also attempted to say what the old API guarantees, but it did not. [15:50:17] old APIs have no guarantees [15:52:46] And opr_Assert isn't really a new version of osi_Assert. osi_* is technically the kernel abstraction layer - bits of it just leaked into userspace over the years. [15:53:02] Right. [15:53:52] The question I am really asking myself right now is, do I care to change this code so that something currently the argument to osi_Assert which *must happen* is not the argument to osi_Assert. [15:54:50] Well, if the behaviour of osi_Assert is changed, everything in tree will need to be changed at the same point. I guess it depends on whether you're going to be in tree by that point or not. [15:55:43] I think this will be in the tree at that point. [15:55:56] So I guess I will leave it for now. [15:59:51] Yeah, doesn't seem much point jumping though hoops to avoid it at the moment. [17:22:32] --- mvitale has left [17:45:59] --- mvitale has become available [18:46:14] --- mvitale has left [20:04:03] --- mvitale has become available [20:04:07] --- mvitale has left [20:12:07] --- Brandon Allbery has left [20:15:23] --- Brandon Allbery has become available