- Wednesday, April 1, 2015

release-team@conference.openafs.org

Wednesday, April 1, 2015< ^ >

Room Configuration

Room Occupants

GMT+0
[10:56:16] shadow@gmail.com/barnowlE5B64A04 leaves the room
[10:56:25] shadow@gmail.com/barnowlE5B64A04 joins the room
[13:19:13] wiesand joins the room
[13:40:40] meffie joins the room
[13:54:06] kaduk joins the room
[14:00:13] <kaduk> $timeofday
[14:00:19] <wiesand> Hi crowd
[14:00:38] <meffie> hello
[14:00:53] <wiesand> I guess this will be another brief one.
[14:01:12] <kaduk> I have another meeting at eleven, anyway.
[14:01:18] <kaduk> Er, "in one hour".
[14:02:40] <wiesand> Fine, let’s go through the agenda.
[14:02:45] <wiesand> Linux 4.0: any news?
[14:03:36] <wiesand> So, no. Gerrit 11809: Ben, caould it possibly do any harm?
[14:04:04] <kaduk> Not anywhere other than FreeBSD.
[14:04:19] <wiesand> Even I got that ;-)
[14:04:44] <wiesand> But on any FBSD system? Presumably older ones?
[14:04:46] <kaduk> I guess if someone was still using gcc maybe it could change things.
[14:04:58] <kaduk> So I should go check when the clang switch happened.
[14:05:54] <wiesand> But then they won’t need 1.6.11.1, and we can fix it for the next release.
[14:06:09] <kaduk> Fair enough.
[14:06:23] <kaduk> FreeBSD 10.0 and newer are using clang, at least.
[14:07:43] <wiesand> The Jeffreys seem not to agree on 11774... any input?
[14:07:54] <Jeffrey Altman> hello
[14:08:04] <wiesand> Hello
[14:09:02] <wiesand> Is jhutz’s assessment in his last comment correct?
[14:09:41] <Jeffrey Altman> the change I referenced as a dependency is a big change and not all of it is required.  One of the smaller bits of the change is adding a function that can reset a ubik server structure to a clean state.
[14:10:06] <Jeffrey Altman> That is the bit that is desired.
[14:10:22] <kaduk> You just want ubeacon_ReinitServer()?
[14:10:27] <Jeffrey Altman> Yes
[14:10:42] kaduk leaves the room
[14:10:42] kaduk joins the room
[14:11:32] <Jeffrey Altman> and the call to it
[14:12:00] <kaduk> That seems to only do anything if !afsconf_UpToDate() (another new function from the named commit)
[14:13:17] <wiesand> Good point.
[14:13:17] <Jeffrey Altman> perhaps.  I haven't had time to read OpenAFS code.
[14:13:31] <Jeffrey Altman> especially 1.6
[14:14:15] <kaduk> Which is ... checking if the KeyFile changed (not even the CellServDB!)?
[14:14:26] <wiesand> We don’t have to sort it out today. It’s not for 1.6.11.1.
[14:14:30] <Jeffrey Altman> and it won't happen this week as I'm at the windows filesystem interop
[14:15:16] <wiesand> I shouldn’t have paster RT #132020 into the agenda. That’s stalled.
[14:15:22] <Jeffrey Altman> the only change that is absolutely required is the first one.   Beyond that as Jeffrey says you end up with added delays.
[14:15:47] <kaduk> I am unhappy at the prospect of needing out-of-tree bits to build an installer usable on win10, but that is not a conversation for here (I will send mail).
[14:15:58] <wiesand> And I guess there’s not much to say about getcwd() either.
[14:17:09] <wiesand> So, unless anyone wants to bring up something coming up for the next release, I think we can move on to 1.8. Ben?
[14:17:26] <meffie> i dont have anything new for 1.6
[14:17:26] <kaduk> We seem to be moving forward, slowly.
[14:18:16] <kaduk> I think Mike's log-rotation changes are next on my review list; I updated the commit moving AFS_PATH_MAX-sized buffers to the heap according to the review comments
[14:18:28] <kaduk> IIRC no one has reviewed akeyconvert yet
[14:19:28] <Jeffrey Altman> I am happy to discuss the windows installer here.
[14:19:36] <kaduk> Jeff has been merging some things that have been sitting for a while; thanks for that.
[14:19:44] <Jeffrey Altman> I've been trying
[14:20:06] <Jeffrey Altman> lots of travel the past few weeks and my dog has been in and out of the vet's office.
[14:20:15] <Jeffrey Altman> many distractions
[14:20:50] <kaduk> I didn't really get anything done last week while I was at the IETF, and I took a sick day on monday since apparently I caught something on the way back.
[14:20:54] <kaduk> So I can sympathize
[14:21:26] <Jeffrey Altman> they ruled out cancer of the prostate yesterday so that is good news.    still no clue what is going on though
[14:22:28] <wiesand> I caught something too...
[14:23:05] <Jeffrey Altman> but to clarify about the windows installers.  its not that someone couldn't write a new installer for windows that could be in tree but I'm not going to do it
[14:23:05] <wiesand> Jeff: sorry to hear that.
[14:23:49] <kaduk> Is it just ("just") a matter of supporting a more recent version of WiX?
[14:24:06] <Jeffrey Altman> The installer is the least of the issues.
[14:24:36] <Jeffrey Altman> EV certs and SHA-256 signatures which are applied by Microsoft to the drivers is the biggest issue
[14:25:09] <Jeffrey Altman> EV certs are $1500 a year
[14:25:19] <Jeffrey Altman> The MSDN subscription is $3500
[14:25:25] <kaduk> Sure.
I know we have to get a SHA-256 cert for signing KfW starting pretty soon, but I don't know if it has to be EV.
[14:25:57] <Jeffrey Altman> Being approved by Microsoft SysQual requires NDAs, insurance and lots of legal agreements
[14:28:05] <Jeffrey Altman> To do all of that on top of the certs necessary for the existing installers is not something I am willing to do.     Just as I am unwilling to do so for OSX.
[14:28:43] <kaduk> I am not trying to ask you to spend thousands of dollars more on openafs than you already are.
[14:29:42] <kaduk> I just want to avoid getting into a state where someone can't clone the tree and figure out how to build an installer (without having to write a new installer), which is where KfW was a few years ago.
[14:30:10] <kaduk> For OpenAFS now (pre-win10), I was in fact able to go do that (with, IIRC, just a couple of general hints from you).
[14:32:37] <Jeffrey Altman> The current installers with separate 32-bit and 64-bit msis and no kerberos dependencies satisfied etc. do build.   The user experience is horrible.   There is no built in Kerberos.   and it won't install without the appropriate signatures.   The Win10 problem is that you can't use the same signatures for Win10 RTM that are currently used for everything else today
[14:33:00] <Jeffrey Altman> That means you need to build one set for Win 7 and one set for Win 10
[14:33:09] <kaduk> *nods*
[14:34:23] <Jeffrey Altman> I have to leave.   You might want to hold off on your response until I send my follow up at the end of this week with the missing bits that I won't get until tomorrow
[14:34:40] <Jeffrey Altman> bye
[14:34:41] <kaduk> Sure.  I have plenty of other backlog to clear which I can do first.
[14:36:58] <wiesand> So the windows state of affairs is just as sad as the OS X one...
[14:37:34] <kaduk> And the linux state of affairs is differently sad.
[14:37:42] <kaduk> So ... everyone move to solaris?
[14:38:54] <wiesand> Yeah.
[14:39:56] <wiesand> I guess we’re done for today?
[14:40:29] <kaduk> seems like it.
[14:41:42] <wiesand> I’ll send minutes later. Will you reply with the 1.8 bits (including the Win installer issue) again?
[14:42:26] <kaduk> I guess so.  There's not much to say about 1.8, and I don't think I want to include the windows installer stuff in the minutes.
[14:43:05] <wiesand> Why not?
[14:43:29] <kaduk> I would like to wait to have discussion on that until Jeffrey can reply with the more information he expects tomorrow.
[14:44:16] <wiesand> Ok.
[14:45:03] <wiesand> Thanks a lot for being here today! Goodbye.
[14:45:23] wiesand leaves the room
[14:46:56] meffie leaves the room
[14:54:07] kaduk leaves the room