[00:13:18] --- manfred has become available [00:14:39] --- dev-zero@jabber.org has left [00:18:34] --- dev-zero@jabber.org has become available [00:28:04] --- manfred has left [00:28:04] --- manfred has become available [00:28:19] --- manfred has left [01:05:34] --- reuteras has left [01:30:00] --- reuteras has become available [01:33:38] --- sxw has become available [02:37:59] --- dragos.tatulea has become available [02:53:34] --- sxw has left [03:26:52] --- sxw has become available [03:42:28] --- SecureEndpoints has left: Replaced by new connection [05:02:27] --- SecureEndpoints has become available [05:02:30] --- sxw has left [05:18:31] --- reuteras has left [05:19:03] --- reuteras has become available [06:15:23] --- matt has become available [07:39:08] --- dmontuori has become available [08:22:25] --- dmontuori has left [08:22:31] --- dmontuori has become available [08:27:57] --- dmontuori has left [08:37:19] --- reuteras has left [08:46:20] --- dmontuori has become available [08:49:41] --- dragos.tatulea has left [10:19:49] --- dev-zero@jabber.org has left [10:35:24] --- Russ has become available [11:31:22] --- dev-zero@jabber.org has become available [11:44:07] --- dmontuori has left [11:44:07] --- dmontuori has become available [11:56:20] --- dmontuori has left [12:18:19] the irix compiler is good, at least, for fussing at prototypes [12:27:17] the digital unix compiler was an excellent torture test [12:29:31] --- Simon Wilkinson has left [12:31:41] 'your code must be a subset of both 4.1 BSD and SVR4' [12:48:14] actually, in this case it's catching places we deviate from prototypes. [12:51:13] that's a good thing, then. [14:08:58] --- dmontuori has become available [15:15:04] --- Simon Wilkinson has become available [15:16:41] shadow: Did you find problems with the IRIX compiler, then? [15:36:38] a problem was found with the function declaration of bcdb_listDumps() [15:38:40] Ah, and I see it's one of those evil compilers that doesn't have an 'inline' option for function definitions. Which perhaps means that my cunning plan for the cache manager needs to be put on hold a little. [15:40:03] If I could get to the point where I could build all of the tree with "treat warning as an error" enabled, I could catch these on Windows as well [15:40:27] Well, I've got a huge pile of patches that should help with that. I just want to test each one before I send it in. [15:40:41] I could put those patches somewhere if others would be interested in doing some of the testing ... [15:41:24] In this case what was noticed by the compiler was that the prototype was not an exact match for the declaration. When Derrick went to fix that he noticed that the first parameter in the function declaration was of the completely wrong type. [15:42:10] On Windows I'm never going to be able to build without warnings until we deal with all of the 32-bit time issues [15:43:10] I have a patch that goes some way to dealing with ctime() issues, at least on Unix, if that's what you're referring to. [15:44:00] We have a lot of assignments back and forth between afs_int32 and time_t [15:44:28] time_t being a 64-bit value on Windows [15:44:35] even on 32-bit systems [15:44:54] assuming you are using a compiler newer than 2002 [15:45:32] Different problem from the one I fixed, then ... [15:45:57] --- sxw has become available [15:47:15] to fix my problem the right way we need to test each assignment of a time_t to a 32-bit time field in an afs structure to ensure it fits and if it doesn't do the right thing. [15:47:37] which probably means inventing a time overflow error [15:47:37] My quilt queue currently has 36 patches in it. I really ought to do something with them ... [15:48:34] Yeh, and getting that error back to somewhere that can do something useful with it could be challenging in some places. [15:49:20] To minimise that challenge, you'd also need to change the 'API' so whenever we're passing time internally, we use a more suitable structure. [15:49:24] I started down this road a few years ago. Discovered it was going to be really hard. Decided it might be easier to deploy a new file system protocol. And walked away [15:50:51] I think you'd need to decide that you didn't care about API stability, and do the kind of thing you require when time gets passed on the wire. [15:51:06] s/you require/you said/ [15:51:51] I figure at some point all of the on disk structures and rpc messages will be changed. When everything is 64-bit time, then I won't care [15:52:19] Windows 7 is going to be fun. [15:52:28] once again browsing \\AFS is broken [15:52:35] I wonder what else broke [15:52:38] Your problem or theirs? [15:52:55] Their if they care about backward compatibility. Mine if they don't [15:53:44] --- mmeffie has left [15:54:33] I will be at Microsoft in ten days to perform Windows 7 file system testing. I will point it out to the engineers while I am there. [15:54:48] Cool. [15:56:10] I wonder if kwc is going to that. Did you ever talk to him (secureendpoints)? [15:58:11] who? [15:58:22] kevin coffman [15:58:24] sorry [15:59:14] I have no idea. I doubt it. I've been going to these events for years. No one from academia has ever attended [15:59:45] ok [16:00:26] There really is no point to attending unless you have a shipping product [16:00:43] and your product runs on Windows [16:01:24] --- mcohan has left [16:01:33] --- mcohan has become available [16:05:09] matt: Are CITI doing a NFSv4 client for Windows, then? [16:05:28] rumor has it that MSFT is paying them [16:05:33] I was told that, and that [16:05:45] Kevin is working on it--I believe it's early stage [16:06:23] Interesting times, indeed ... [16:07:42] shadow: parm.i386_obsd44.h lost somewhere [16:08:07] (That'll be the same MSFT that just announced its shedding 5% of its workforce, after an 11% drop in profits) [16:08:09] erm, param [16:08:38] when you have open source developers, who needs staff. [16:08:55] its not like you have to pay them a real salary or anything [16:09:10] Well, I think that was Apple's approach when they built Mac OS X originally. [16:09:14] send them some free devtools and some pot [16:09:33] how do you get the...free devtools? [16:09:48] ask Peter [16:10:00] think he quit [16:10:02] I pay for mine [16:20:08] shadow: few more things for openbsd4.x--got time to wait for them? [16:20:34] he is offline [16:20:53] does that say anything about 1.5.57? [16:21:00] eg, tomorrow? late tonight? [16:21:19] and if you look at cvs you will see a tag was already applied [16:21:36] I don't think anyone is waiting on openbsd, anyway. oh well [16:21:55] got freebsd 70 and 71 at least building before that [16:22:05] I don't think it's worth worrying about catching the 1.5 release train. [16:22:23] I've got stacks of disconnected stuff that didn't quite make it (although the stuff that's there is now usable) [16:22:25] there will be another 1.5.x release next month [16:23:23] k [16:24:17] its not like we are shipping binaries. you can always tell people that are building from source to use the cvs head as of such and such a date [16:24:30] or the 1_5_x branch [16:24:41] right, trying to head off some of the freebsd emails--think we did that [16:31:14] If there is something important, we can always retag [16:31:55] If its not, just create tickets for the items and they will be addressed before the next release. [16:32:08] k [16:39:42] ug. 1_5_x openbsd is missing changes that are on 1_4_x [16:40:13] from how long ago? [16:40:23] looks like over a year [16:40:41] you can thank a former gatekeeper [16:40:45] well... [16:41:23] they might be missing from the head as well [16:42:41] how crazy would it be to just resync client with what 1.4 has--since that works atm [16:42:52] (not tonight) [16:43:03] If you would like to have some idea of where Network Identity Manager is headed, take a look at http://www.secure-endpoints.com/netidmgr/roadmap.html. [16:43:37] I've added a number a screen shots and highlights of the new functionality [16:44:02] matt: not sure what you mean by that, but I don't think that the 1.5 unix client looks very much like the 1.4 one any more. You and I have seen to that ... [16:44:48] ... amongst many others. [16:45:00] well, insofar as those changes would apply. I also find it odd that openbsd private functions all have "nbsd" in the name, etc [16:45:36] I guess that's partly illogical. will need to merge even that stuff, I guess [16:46:18] I've given up on "that's odd" in the CM. I'm starting to get rid of the bits that bug me, and I'll probably keep chipping away, as it seems to be where I'm spending most of my OpenAFS coding time ATM. [16:46:34] we have a proposal to sync the 1.4 servers with the 1.5 clients. why not a proposal to sync the 1.4 clients with the 1.5 servers [16:47:16] I didn't mean to discard 1.5 changes, sorry if it sounded like that [16:47:28] SecureEndpoints: Go for the worst of both worlds, you mean? [16:47:35] absolutely [16:48:03] and while we are at it we can throw some 1.2 into the mix [16:48:16] lets bring back the win95 support [16:48:48] Which reminds me. Can we kill CForeign? [16:49:03] (the DFS support in the Unix CM) [16:50:09] as opposed to the Dfs support in the Windows client :-) [16:50:35] Yes. Isn't it great that that acronym has so few meanings ... [16:50:54] Tom said psu just turned off DFS. Don't know if that means, somebody still hasn't [16:51:06] ANL might [16:51:15] and IBM absolutely does [16:51:32] --- abo has left [16:51:33] --- mcohan has left [16:51:34] So that's probably a 'no', then. [16:51:36] and we aren't pulling the rug out from under them quite yet [16:51:53] but I will add it to the list of things to do when the lawsuits start [16:52:15] --- abo has become available [16:52:22] --- mcohan has become available [16:54:38] sxw: that's odd: I'm glad someone is; there seems to be a lot brush to be cleared [16:55:17] --- matt has left [17:02:52] --- dmontuori has left [17:22:08] > Can we kill CForeign? I wish you wouldn't. hostafs depends on that functionality [17:25:11] Another two disconnected bugs bite the dust... [17:25:15] Well, OK, not quite "depends". It is capable of pretending to be a translator, and does so by default, in order to trick the cache manager into handling permissions in a more useful way. [17:26:03] I doubt we'll get rid of it from what Secure Endpoints says about DFS. But it is a fair chunk of code in the Unix CM that I would imagine gets ~no testing at all. [17:26:07] sxw: you are making me feel bad. I can't knock off windows bugs at this rate any longer [17:26:19] Not even in the IFS? [17:27:02] I have to find the bugs first in order to be able to fix them and that take times [17:27:14] plus I'm working on so many other things [17:27:47] Yeh. I spent a good while testing the disconnected code, reading the disconnected code, and writing a bug list. So, I've got a fair amount to work through. [17:28:23] netidmgr is taking a fair bit of my time these days. as is testing fixes for msft [17:28:39] I guess you've got to go where the money is! [17:28:49] I didn't say I got paid for any of that [17:28:59] its just where my time is being spent [17:29:16] if someone knows where the money is located I will go head in that direction [17:29:38] --- stevenjenkins has left [17:30:13] apparently its not in the banks, and its not in the hands of the VCs nor in the hands of the hospitals or the real estate companies [17:30:14] I did hear something about the end of a rainbow. [17:30:27] What's there is probably worth more than sterling at the moment. [17:31:51] Looking back on it I probably should have accepted that job I was offered last Summer but the consequences of accepting it were just too great. [17:32:57] --- stevenjenkins has become available [17:33:02] Hindsight is a wonderful thing... But, if it was a bad idea then, it's probably still a bad idea now. [17:33:16] that depends on how much I like Asanka [17:33:56] sending him and his wife back to a war zone is not exactly my idea of a good time [17:34:15] that doesn't sound great, no. [17:47:44] The money is in the hands of people who are holding on to it because they think the economy is declining. [17:49:00] and because they don't have as much of it today on paper as they did six months ago [17:50:50] or because they don't have as much credit today as they did six months ago. [18:56:51] --- mcohan has left [18:57:14] --- mcohan has become available [20:19:04] --- dev-zero@jabber.org has left [23:22:03] --- reuteras has become available