[00:53:09] --- kaj has become available [02:20:12] --- Simon Wilkinson has left [03:48:09] --- Simon Wilkinson has become available [03:50:13] --- Simon Wilkinson has left [03:56:56] --- Simon Wilkinson has become available [04:48:54] --- Simon Wilkinson has left [05:00:51] --- mdionne has become available [05:08:47] --- Kelli Ireland has become available [05:12:41] --- mdionne has left [05:53:09] --- matt has become available [06:02:19] --- ktdreyer has become available [06:03:47] --- ktdreyer has left [06:05:04] --- haba has become available [06:16:09] --- jaltman has left: Disconnected [06:19:17] --- Kelli Ireland has left [06:21:31] --- matt has left [06:31:19] --- haba has left [06:40:42] --- mdionne has become available [06:41:05] --- sanket has become available [06:42:09] --- Russ has become available [06:51:20] --- Kelli Ireland has become available [06:54:24] Simon: Yeah, that's what I figured when I thought about it more. [06:54:28] --- jaltman has become available [06:56:19] --- sanket has left [06:57:35] --- snktagarwal has become available [06:58:41] --- reuteras has left [07:01:44] --- deason has become available [07:06:05] --- snktagarwal has left [07:19:01] --- haba has become available [07:26:32] --- matt has become available [07:36:31] --- kaj has left [07:40:27] --- Derrick Brashear has become available [07:45:41] --- mmeffie has left [07:46:37] --- mmeffie has become available [08:21:37] * Russ deletes the careful AFS documentation of the difference between personal computers and mainframes and the discussion of the two basic types of networks: networking personal workstations and connecting terminals to a mainframe. [08:22:06] --- sanket has become available [08:22:42] hah [08:55:40] --- sanket has left [09:12:18] --- haba has left [09:12:41] --- haba has become available [09:55:57] --- sanket has become available [10:00:46] Can someone help me rip data off struct uio. I am trying the following: in the function afs_UFSRead, the variable tuio represents the transfer from cache data to kernel space. My present task is to make some transformations to the data held by this tuio variable. [10:26:09] --- haba has left [10:26:43] --- haba has become available [10:48:57] --- sanket has left [10:56:38] --- haba has left [10:57:33] --- Derrick Brashear has left [11:00:27] --- matt has left [11:00:41] --- jaltman has left: Disconnected [11:02:14] --- deason has left [12:12:37] --- deason has become available [12:19:24] --- jaltman has become available [12:20:01] --- kaj has become available [12:30:02] --- kaj has left [12:30:08] --- kaj has become available [12:33:43] --- kaj has left [12:36:26] --- kaj has become available [12:38:37] --- mmeffie has left [12:44:32] --- JSund has become available [12:46:18] hi Jonas [12:46:39] I haven't been able to stay logged into IRC recently due to my schedule. [12:47:03] hi [12:47:21] a couple of days ago you asked about testing frameworks. Chris Clausen pointed you at C TAP. Does it look reasonable to you? [12:52:02] I didn't find any documentation on how to write the actual tests when I took a look at the site, so I looked more at Check instead [12:54:20] Also, I've used JUnit before, so I think I would be more familiar with some kind of xUnit framework. [12:56:05] The libtap/basic.h header file is basically the documentation. It could stand to have better documentation, but basically you call plan() with the number of tests you're running and then use the ok() and is_*() functions. [12:56:27] I don't think xUnit is a good idea for testing C code; you end up spawning lots and lots of separate programs if you're using an assertion framework. [12:56:33] Either that or you can only report the first failure. [12:56:48] It would work with C++, of course. [12:57:22] TAP is pretty simple -- take a look at any of the tests in rra-c-util (http://www.eyrie.org/~eagle/software/rra-c-util/) for some good examples. [12:58:00] --- Kevin Sumner has left [12:58:01] TAP looks a lot like xUnit except it uses textual output instead of assertion tests. [12:58:52] --- Kevin Sumner has become available [12:59:52] Okay, I'll take a look at the tests for rra-c-util. [13:02:24] Any opinions on using CppTest for testing a C library by the way? [13:05:29] Well, in the AFS context, the test suite on platforms other than Windows should ideally not require C++, IMO. We don't require C++ to build AFS now, and it adds a new programming language and set of infrastructure to our profile on UNIX. [13:05:42] --- kaj has left [13:05:43] --- kaj has become available [13:06:09] Yeah, as I thought then. [13:08:43] Check seems to run each test in a separate fork() unless that feature is disabled for debugging specific crashes. [13:12:30] That's probably how it works around the lack of catching assertions in C. [13:13:18] anything that relies on fork() will not work on Windows. [13:13:24] I believe check can be kicked into not using fork(). [13:13:46] Yeah, but then the entire run will crash if you run into a segmentation fault [13:14:27] for this project I think it is very important that the test framework be usable on Windows so it can be run against Microsoft's implementation. This is needed to verify the actual behavior matches the documentation. [13:14:45] Yeah, then I guess Check is out. [13:19:31] The runtests framework in C TAP Harness forks and opens pipes to the test children, but that's only to collect results; I suspect that one could collect results in other ways on Windows (possibly even by just using the Perl code it was modeled after). [13:28:37] Do you recommend that I put the C TAP Harness files inside the project and distribute them in the repository? [13:29:18] I think I saw some notice about that when I scrolled through the web site. [13:29:42] Yes, that's the recommended way to use it. [13:43:45] --- mdionne has left [14:00:59] --- mdionne has become available [14:06:39] if "make clean" removes all the stuff from a build, "make lean" should strip the binaries [14:09:47] And make lee should move all your built binaries out of the wind. [14:11:26] ... or at least to the other side of the boat [14:42:36] --- mdionne has left [14:50:43] 'make lean' should be an alias to mkfs.ext3 (i.e., it removes FAT...) [14:50:51] yes, I'm both late and lame.. [14:59:08] --- Kelli Ireland has left [15:12:24] --- deason has left [15:21:16] --- kaj has left [15:30:08] --- jaltman has left: Disconnected [15:31:48] --- jhutz@jis.mit.edu/owl has left: Lost connection [15:39:17] --- Russ has left: Disconnected [15:44:18] --- jaltman has become available [15:49:50] --- jaltman has left: Disconnected [15:49:58] --- JSund has left [19:18:05] --- Kelli Ireland has become available [19:19:24] --- Russ has become available [19:25:52] --- haba has become available [19:40:43] --- jaltman has become available [19:55:44] --- Russ has left: Replaced by new connection [19:55:45] --- Russ has become available [20:02:23] --- haba has left [20:02:41] --- haba has become available [20:16:24] --- haba has left [20:16:25] --- haba has become available [20:21:32] --- Kelli Ireland has left [20:31:24] --- haba has left [20:31:24] --- haba has become available [20:38:23] --- haba has left [20:47:58] --- haba has become available [20:54:27] given that pioctl could be used on non-afs files, that would be fine btw (pioctl vs afs) [20:56:14] I'm just not entirely sure if it would be worth the code churn, as it would not necessarily be available on all our platforms. [20:58:59] --- haba has left [21:02:49] --- Born Fool has become available [21:13:27] there's code that does it already unless we just kileld it [21:22:28] Oh. I guess I should look harder. [21:45:33] --- kaj has become available [21:45:47] --- kaj has left [22:16:22] --- reuteras has become available [22:23:39] --- Russ has left: Disconnected [23:17:13] --- Born Fool has left