02:41:22  * jgiquit (Quit: jgi)
03:25:30  * jgijoined
03:48:12  * jgiquit (Quit: jgi)
04:35:44  * jgijoined
04:45:22  * jgiquit (Quit: jgi)
04:52:19  * jgijoined
04:54:46  * jgiquit (Client Quit)
05:16:51  * jgijoined
06:36:48  * jgiquit (Quit: jgi)
09:32:12  <jbergstroem>https://github.com/creationix/nvm/issues/906
09:32:14  <jbergstroem>i wonder what happened
10:35:55  <jbergstroem>found it: https://github.com/nodejs/build/issues/258
10:36:42  <jbergstroem>rvagg: is it possible to recreate them? nvm will cry if we don't (it looks for xz tarballs over gz ones if possible)
10:58:05  <rvagg> I need to get in to the 32-bit centos release machine and I can't figure it out. I've tried using stuff from ibm-ppc.txt but to no avail, can you show me an ssh config for this machine or something?
10:58:10  <rvagg>jbergstroem: ^
10:58:34  <jbergstroem>ibm-ppc.txt
10:59:11  <jbergstroem>should give you rsa. password is in there too (i mistook it for some kind of root password but its actually the key password)
10:59:23  <jbergstroem>rvagg: sorry -- secrets/build/ibm-ppc.txt
11:00:14  <rvagg>jbergstroem: right but what can I do with it? I can't get in to the release machine, are you sure we're reusing build credentials to get in to release machines?
11:00:34  <jbergstroem>rvagg: yeah, that's the key i'm using to log in to it.
11:00:37  <rvagg>I'm using
11:00:41  <rvagg>username `ubuntu`?
11:01:03  <jbergstroem>hosts here: https://github.com/nodejs/build/tree/master/setup/centos5
11:01:16  <jbergstroem>.142 -- isn't that the 64-bit release ?
11:02:20  <rvagg>yeah, sorry,, is that the 32-bit one?
11:02:48  <rvagg>grrr got it now
11:02:55  <jbergstroem>\o/
11:02:57  <rvagg>sorry, I see `root` in the setup readme
11:03:03  <jbergstroem>can i leave baking the xz in your hands?
11:03:15  <rvagg>also, password protected ssh keys
11:03:17  <rvagg>grr
11:03:24  <rvagg>yeah, I'll deal with xz while I'm at this
11:03:30  <jbergstroem>great, thanks.
11:27:41  <rvagg>my attempt to be tricky with compiling Node with the default compiler on Centos5 for 0.12 have failed because if I don't use devtoolset-2 I don't get a an updated python https://ci.nodejs.org/job/iojs+release/289/nodes=centos5-release-32/console
11:32:07  <rvagg>ah, /home/iojs/bin has it
11:52:06  <rvagg>I believe we are pretty much ready for an RC.1 of v0.12.8 https://nodejs.org/download/nightly/v0.12.8-nightly20151118a6d256be0b/
11:52:15  <rvagg>just the SHASUMS hoo haa needed now
17:00:30  * jgijoined
17:11:55  * jgiquit (Quit: jgi)
17:43:31  * orangemochajoined
17:43:59  * jgijoined
17:45:30  * orangemocha_quit (Ping timeout: 240 seconds)
19:27:23  <jbergstroem>rvagg: are the sunos builds from smartos13?
19:27:42  <jbergstroem>jgi: did you get a chance to do any testing yesterday?
19:28:09  <jgi>jbergstroem: yes, I was actually looking for you to give you an update and saw you were away
19:28:26  <jgi>jbergstroem: I’m currently writing up a summary of the problem
19:28:30  <jgi>jbergstroem: it’s a bit complicated :)
19:28:35  <jbergstroem>jgi: great!
19:28:51  <jbergstroem>jgi: btw feel free to write to me regardless if im away or not; i usually get notifications
19:29:00  <jgi>jbergstroem: great, noted!
19:32:36  <jgi>jbergstroem: the conclusion so far is: smartos binaries downloaded from nodejs.org have been working by accident for a long time on a lot of SmartOS machines, and smartos users should install nodejs from pkgsrc. Solaris users should not run binaries built on SmartOS machines.
19:33:30  <jbergstroem>jgi: how up to date is pkgsrc binaries? there's no concept of backports, right?
19:35:43  <jgi>jbergstroem: newest version available in pkgsrc 2014Q4 is v4.2.1, I don’t know why v4.2.2 is not available, checking now
19:35:49  <jgi>jbergstroem: what do you mean by “backports”?
19:36:31  <jbergstroem>jgi: just if security stuff is backported to for instance 2013Q4
19:36:55  <jbergstroem>jgi: reading between your lines it seems like we should stop releasing for smartos and point to pkgsrc -- if so we need to make sure users get security updates.
19:40:51  <jgi>jbergstroem: you mean are nodejs releases that fix security issues made available in older pkgsrc repositories?
19:41:00  <jbergstroem>jgi: yeah
19:41:13  <jgi>jbergstroem: that’s a good question, but I would think that yes, they would, at least I hope so
19:41:39  <jgi>jbergstroem: I’m not saying we should stop making smartos releases available from nodejs.org, because today I don’t know how many people download them?
19:42:06  <jgi>s/?/./ :)
19:42:12  <jbergstroem>jgi: ok, but if there's no guarantee they will work how to we handle a scenario when they stop :)
19:44:14  <jgi>jbergstroem: there are many moving parts. Maybe we (Joyent) can make these binaries work in newer images, and over time more images would have proper (as in, not by accident) support for these binaries. I’m writing that down in more details so we can determine what’s the best way forward.
19:44:44  <jbergstroem>jgi: that'd be awesome. just throwing it out there but how would a binary built on solaris play?
19:45:15  <jgi>jbergstroem: what do you mean?
19:45:35  <jbergstroem>jgi: if a nodejs built on solaris would work on smartos
19:49:07  <jgi>jbergstroem: it might work, like v0.10.x and v0.12.x node smartos binaries work on Solaris.
19:49:27  <jgi>jbergstroem: however there are a few problems to consider that make this not a viable solution:
19:49:32  <jbergstroem>jgi: speaking of 0.12; there's a test release here: https://nodejs.org/download/nightly/v0.12.8-nightly20151118a6d256be0b/
19:49:45  <jbergstroem>jgi: haven't confirmed but afaik that should be built on smartos13
19:49:54  <jgi>1) the Solaris toolchain will change, and may become incompatible with various SmartOS runtimes.
19:50:38  <jgi>2) the assumptions that are made at compile time to change the behavior of node’s native layer may break node when it runs on SmartOS.
19:50:57  <jgi>for instance, #ifdef SOLARIS doSomethingThatBreaksOnSmartOS #endif
19:51:26  <jgi>in general, one should always build a binary on the exact oldest platform on which that binary must be able to run
19:51:57  <jgi>running such a binary on a newer platform is guaranteed to work (by the C and C++ ABI), but not the other way around
19:52:02  <jgi>and also not accross different platforms
20:55:37  <rvagg>jbergstroem, jgi: yes, 0.12 builds are coming out of that smartos13 box, v4+ are done on the smartos15. This is controlled by the pre-1-release and post-1-release labels in Jenkins.
20:55:44  <jbergstroem>rvagg: awesome
20:56:08  <rvagg>Going to cut the RC.1 today.
20:56:20  <jbergstroem>jgi: ok, but if we were to compile on an older solaris? i contacted two vm providers just to get a test trial
21:01:32  <jbergstroem>rvagg: should we perhaps port the xz stuff to 0.12.x?
21:36:51  <rvagg>jbergstroem: I explicitly left it out when porting, I figured it'd be best to be as close as possible to 0.12 release style, the directories should look pretty much the same as the older ones
21:37:00  <jbergstroem>rvagg: ok, cool.
21:57:52  <jgi>jbergstroem: if we were to compile on older solaris and use provide these binaries for Solaris _and_ SmartOS, or just for Solaris?
22:00:19  <jbergstroem>jgi: both if possible of course?
22:00:41  <jgi>jbergstroem: no, we should build Solaris binaries on Solaris, and SmartOS binaries on SmartOS
22:00:54  <jbergstroem>jgi: sure
22:01:19  <jgi>jbergstroem: the same way that if we had OpenBSD binaries, they might work on NetBSD, but we would still build NetBSD binaries on NetBSD
22:02:33  <jbergstroem>jgi: i totally get that but I don't understand how much fork in terms of libc/libstdc illumos has got on solaris which is why i'd prefer asking
22:02:45  <jbergstroem>jgi: i do have that experience with open and netbsd though.
22:04:11  <jgi>jbergstroem: I don’t really know to be honest, but there are differences, and unless there’s a clear strategy to make these systems ABI compatible, we should assume they’re not the same
22:04:24  <jgi>jbergstroem: or rather, that they’re note even remotely close
22:05:47  <jbergstroem>jgi: ok; then we need solaris builders if we're to provide support.
22:06:21  <jgi>jbergstroem: yes, and it’s great that you’ve looked into cloud providers that would help us do that
22:06:26  <jgi>jbergstroem: so thank you very much :)
22:06:39  <jgi>jbergstroem: and I hope I’ll be able to help you get that going too
22:07:38  <jgi>jbergstroem: also, v4.2.2 will be in SmartOS’ 2014Q4 pkgsrc tonight
22:07:50  <jbergstroem>jgi: my suggestion would then be that we introduce -smartos when we have solaris builds available in -sunos
22:08:38  <jgi>jbergstroem: yes, but like I said we need to evaluate the impact on nvm users
22:08:53  <jgi>jbergstroem: and users of other similar versions managers
22:39:01  <jgi>jbergstroem: what I managed to write down so far: https://gist.github.com/misterdjules/eae9ec70dc1d91fb8dd1
22:39:34  <jgi>jbergstroem: it’s probably still a bit confusing to read, I’ll let you know when I’m done, and then I would appreciate any feedback to make that easy to read so that we can discuss next steps :)