Switch to DuckDuckGo Search
   April 20, 2020  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | >

NOTICE: This channel is no longer actively logged.

Toggle Join/Part | bottom
[00:00:14] *** ghost1 <ghost1!~ghost1@c-73-245-52-228.hsd1.fl.comcast.net> has quit IRC (Quit: ghost1)
[00:05:48] *** ypankov <ypankov!~ypankov@> has joined #illumos
[00:25:28] *** ldepandis <ldepandis!~ldepandis@unaffiliated/ldepandis> has quit IRC (Read error: Connection reset by peer)
[00:30:54] *** jellydonut <jellydonut!~quassel@s91904423.blix.com> has quit IRC (Quit: jellydonut)
[00:32:10] *** jellydonut <jellydonut!~quassel@s91904423.blix.com> has joined #illumos
[00:39:28] *** phyre <phyre!~phyre___@> has quit IRC (Remote host closed the connection)
[01:04:34] <richlowe> ypankov: there's a branch in github.com/richlowe/illumos-gate that removes translations from cw
[01:04:36] <am11> rmustacc: i added note based on your earlier comment https://github.com/dotnet/runtime/pull/35173#discussion_r410863333. hope that is the correct expectation. :)
[01:05:29] <richlowe> rmustacc: were you happy with my updates based on your review comments?
[01:07:34] * toastersonerson1 wants /reconfigure for nics....
[01:54:33] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[01:54:48] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has joined #illumos
[02:02:43] <andyf> @tsoome - are you looking at bug 12469?
[02:05:15] <andyf> It makes it impossible to install driver/network/mlxcx on omnios
[02:05:37] <andyf> trivial to fix, but if you're already on it...
[02:06:42] <ypankov> andyf: do we need those "legacy" lines at all?
[02:06:59] <andyf> not for me.. possibly Tribblix uses them to build the SVR4 packages
[02:07:00] <ypankov> Yes, I understand it should be fixed, just a generic question
[02:07:05] <ypankov> ah I see
[02:08:42] <ypankov> https://pastebin.com/R2K1n8ZA <- would anyone find printing the controlling terminal in ptree useful? I just added to show the problem with tmux/w
[03:17:53] <richlowe> the legacy lines are used to fake install information for 3rd party svr4 packages too
[03:23:22] *** varna_ <varna_!~varna@> has joined #illumos
[03:25:56] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has quit IRC (Quit: ZNC 1.7.5 - https://znc.in)
[03:26:36] *** varna <varna!~varna@> has quit IRC (Ping timeout: 256 seconds)
[05:14:39] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[05:23:23] <LeftWing> toastersonerson1: You may need to remove any existing instances of vioif from /etc/path_to_inst when you create the image?
[05:24:05] <LeftWing> On next boot I would them expect them to enumerate again
[05:33:47] <Smithx10> jbk: I wonder if there would be any possibility to have a illumos and solaris distinction like there is in Go?
[05:33:52] <Smithx10> https://twitter.com/jasonbking/status/1252036851860373504?s=20
[05:41:17] <LeftWing> I would certainly encourage using a distinct illumos identity in there as soon as possible
[05:41:59] <LeftWing> It will inevitably be necessary even if it doesn't seem to be now, and it helps to clarify that testing on Oracle Solaris is not required
[05:42:43] <LeftWing> I feel that every time we haven't done it straight off the bat we have regretted it later and had to do more work (e.g., see Go, Rust, etc)
[05:43:55] <rmustacc> richlowe: I'll try to cycle back to that again, soon. Sorry, bit behind on reviews.
[05:44:26] <rmustacc> am11: It's pretty close, though probably more of a thing via sysconf or similar rather than a wrapper.
[06:12:11] <Smithx10> rmustacc: I saw your SMF post, do you know if SMF allows for running a hook on state change?
[06:13:04] <Smithx10> The scenario being that I'd want to register a running service into a service catalog etc. Not sure if that already exists
[06:14:23] <Smithx10> I guess you'd do handle that in the start / stop methods .
[06:24:35] <rmustacc> There are system events.
[06:25:26] <rmustacc> Though I think you're probably better off thinking about building that into the service/application itself, tbh.
[06:26:53] <rmustacc> That way you can tie into graceful restarts and all that fun stuff.
[06:31:45] <Smithx10> yea... I was just thinking about the Containerpilot project and how it kind of just turned into an init system that registered services
[06:32:06] <Smithx10> but I believe most of those Service Catalogs have a wrapper for forking the process anyway
[06:32:47] <Smithx10> and probably better to have them listen to system events if those applications were to be registered into something.
[06:33:15] <LeftWing> I vaguely recall that there is an API for subscribing to changes in the SCF database
[06:33:23] <LeftWing> That the SMF ipfilter support uses
[06:33:54] <LeftWing> Not sure if it is a stable public API or not
[06:34:06] <jbk> i don't think it is
[06:35:31] <LeftWing> Ugh I guess it is not
[06:35:38] <LeftWing> _scf_notify_wait() and such
[06:35:49] <Smithx10> I know systemd has things like ExecStartPre and ExecStopPost or whatever
[06:36:25] <LeftWing> So one thing that you can do already is have a second SMF service that depends on the first one starting up
[06:36:38] <LeftWing> And which is configured to restart or stop when the other one restarts or stops
[06:37:01] <LeftWing> I believe Isaac used that in some of the Prometheus configuration in Manta
[06:38:13] <LeftWing> https://github.com/joyent/binder/blob/master/smf/manifests/metric-ports-updater.xml.in#L20-L25
[06:38:28] <LeftWing> I believe restart_on= makes that happen
[06:40:28] <LeftWing> https://illumos.org/man/5/smf#dependencies
[06:41:22] <Smithx10> yea, it was just an idea... but I think rmustacc is right about doing it outside of SMF
[06:42:21] <Smithx10> Not sure tho if supporting registration into a shared catalog might be a thing that should just happen now a days
[06:42:44] <Smithx10> optionally*
[06:42:49] <Smithx10> but its very complicated
[06:42:50] <Smithx10> lol
[06:42:57] <LeftWing> The combination of grouping=optional_all and restart_on=refresh and the transient duration will effectively produce a script that will start up after the dependent service, and be restarted any time it is brought online, fails, is restarted, etc
[06:43:35] <LeftWing> Which is what we wanted for the metric ports registration -- any time anything changed it was supposed to restart and update the registration
[06:43:40] <Smithx10> LeftWing: that's definitely something im gonna save in my notes
[06:43:56] <Smithx10> useful
[07:15:21] *** gitomat <gitomat!~nodebot@> has quit IRC (Remote host closed the connection)
[07:15:31] *** gitomat <gitomat!~nodebot@> has joined #illumos
[07:30:59] *** BOKALDO <BOKALDO!~BOKALDO@> has joined #illumos
[08:01:11] <tsoome> andyf um no, I do not know why it was set for me:D
[08:01:40] <tsoome> if you can pick it up, it would be nice:)
[08:06:58] <LeftWing> andyf: Is there something we should add to the packaging checks to avoid mistakes in future?
[08:18:06] *** tsoome_ <tsoome_!~tsoome@> has quit IRC (Read error: Connection reset by peer)
[08:21:49] *** tsoome <tsoome!~tsoome@7129-0555-394f-640e-2f80-4a40-07d0-2001.sta.estpak.ee> has quit IRC (Quit: This computer has gone to sleep)
[08:31:30] *** phyre <phyre!~phyre___@> has joined #illumos
[08:52:15] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has quit IRC (Remote host closed the connection)
[08:52:42] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has joined #illumos
[09:04:58] <Agnar> moin
[09:09:29] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[09:27:24] *** andy_js <andy_js!~andy@> has joined #illumos
[09:28:40] *** jimklimov <jimklimov!~jimklimov@ip-86-49-254-26.net.upcbroadband.cz> has joined #illumos
[09:30:39] *** tsoome <tsoome!~tsoome@> has joined #illumos
[09:31:18] *** nde <nde!uid414739@gateway/web/irccloud.com/x-eorwrbzwvtgbbimf> has joined #illumos
[09:46:33] <tsoome> fun fact from sparc build; building cmd/halt with gcc 4.4.4, the libzfs and libgen are needed, but with gcc 7, they are not needed.
[09:47:36] <tsoome> gcc 7 does optimize fastboot bits out for sparc.
[09:48:16] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[09:57:00] <Agnar> tsoome: hahaha
[10:15:41] <sjorge> Smithx10 I guess you want to talk to am11 to get solaris/illumos distinction in dotnet, as they are doing most (all?) of the work.
[10:23:08] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has joined #illumos
[10:25:36] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[10:27:51] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has quit IRC (Ping timeout: 256 seconds)
[10:37:02] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has joined #illumos
[10:56:20] *** tsoome <tsoome!~tsoome@> has quit IRC (Quit: tsoome)
[11:17:48] *** amrfrsh <amrfrsh!~Thunderbi@> has quit IRC (Quit: amrfrsh)
[12:03:49] *** dlg_ <dlg_!~dlg@toy.eait.uq.edu.au> has joined #illumos
[12:04:52] *** khng300_ <khng300_!~khng300@unaffiliated/khng300> has joined #illumos
[12:05:08] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 265 seconds)
[12:05:20] *** khng300 <khng300!~khng300@unaffiliated/khng300> has quit IRC (Ping timeout: 256 seconds)
[12:05:55] *** khng300_ is now known as khng300
[12:26:17] <am11> sjorge: there is something called RID - runtime identifier, which we would need to coin for dotnet when we get there. that would require capturing these ancestral details in the form of a tree: https://github.com/dotnet/runtime/blob/aa5b204/src/libraries/pkg/Microsoft.NETCore.Platforms/runtime.compatibility.json
[12:28:15] <am11> usually that is an interesting conversation to convince masses _what makes sense_, as naming is a hard.. :p
[12:35:59] <am11> i was thinking the inheritance will look something like [rank: 0] unix <- sunos, [rank: 1] sunos <- solaris, [rank:1] sunos <- illumos, [rank: 2] solaris <- solaris-x64, [rank: 2] illumos <- smartos [rank: 3] solaris-x64 <- solaris11-x64, [rank: 3] smartos <- smartos-x64, [rank: 4] smartos-x64 <- smartos2019-x64
[12:36:18] <am11> (and since so forth
[12:36:21] *** dlg_ is now known as dlg
[12:37:44] *** amrfrsh <amrfrsh!~Thunderbi@> has joined #illumos
[12:38:16] <am11> however this RID is not strictly to capture binary compatibility, it is just for hinting the tools / editors about the possible fallback path (with its own share of caveats for sure..)
[12:45:40] *** lgtaube <lgtaube!~lgt@213-67-21-71-no84.tbcn.telia.com> has joined #illumos
[12:48:18] *** amrfrsh <amrfrsh!~Thunderbi@> has quit IRC (Quit: amrfrsh)
[13:33:49] *** tsoome <tsoome!~tsoome@> has joined #illumos
[13:44:44] *** tsoome <tsoome!~tsoome@> has quit IRC (Quit: tsoome)
[13:48:15] *** tsoome <tsoome!~tsoome@> has joined #illumos
[13:55:34] *** ldepandis <ldepandis!~ldepandis@unaffiliated/ldepandis> has joined #illumos
[13:55:56] *** tsoome <tsoome!~tsoome@> has quit IRC (Quit: tsoome)
[14:01:57] *** tsoome <tsoome!~tsoome@> has joined #illumos
[14:02:54] *** tsoome <tsoome!~tsoome@> has quit IRC (Client Quit)
[14:11:40] *** phyre <phyre!~phyre___@> has quit IRC (Remote host closed the connection)
[14:41:27] *** amrfrsh <amrfrsh!~Thunderbi@> has joined #illumos
[14:53:11] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has joined #illumos
[14:59:27] *** amrfrsh <amrfrsh!~Thunderbi@> has quit IRC (Quit: amrfrsh)
[15:08:14] <am11> is there a way to define `char16_t` without `std=c++11`?
[15:09:02] <am11> Linux has a non-posix `uchar.h` header for that purpose, macOS, smartOS, FreeBSD etc.do not.
[15:10:41] <toastersonerson1> how is it solved on macOS then?
[15:13:31] <am11> a test suite for dotnet/runtime added dependency on that header few weeks ago, it is currently not in the CI (unintentionally), and broken on non-Linux OS (also unintentionally). i asked the maintainers about it: https://github.com/dotnet/runtime/pull/1313#discussion_r410748432
[15:14:56] <am11> i was thinking about somehow workaround it locally to be able to run this primary test suite (which exercises their platform abstraction layer, multi-threading, thread sequencing, semaphores etc.)
[15:15:58] <am11> if we pass all of PAL suite, we can enter the codegen realm. NetBSD had it all passing in 2017/2018, but it required some additional work at that time.
[15:17:16] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[15:18:35] <am11> currently we have no idea how reliable is the solaris support, and im blocked by this seemingly little char16_t thingy (which all non-Linux OS rightfully _only_ define in C++11+ context https://github.com/illumos/illumos-gate/blob/4e0c5eff9af325c80994e9527b7cb8b3a1ffd1d4/usr/src/boot/sys/sys/_types.h#L92)
[15:19:38] <toastersonerson1> well why not use c++11 if they are using it? also if they are not testing they are breaking things sooo.
[15:24:03] <am11> in PAL layer of dotnet runtime, they undef all the STDCPP functions and redefine them uniformly for all supported OSes. it is a large codebase that does it, and their wish is to remove lots of those redundancies in the future, but requires some additional work especially in EH area:
[15:24:03] <am11> https://github.com/dotnet/runtime/issues/32144#issuecomment-585771406 (they need exceptions to go though the runtime which C++11 std *probably* does not support).
[15:26:41] <am11> therefore, some areas of coreclr which do not use std=c++11 are non-trival to switch over to c++11 (but doable). there are some other areas in coreclr, which use stdc++11.
[15:32:24] <am11> src/installer/corehost use c++11, src/libraries and src/mono are plain ol' gnu99.
[15:34:37] *** tsoome_ <tsoome_!~tsoome@> has joined #illumos
[15:35:40] <am11> src/coreclr is mixed, some parts use stl, others use redefined standard functioned from pal
[15:35:54] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 240 seconds)
[15:35:54] *** tsoome_ is now known as tsoome
[15:39:54] *** varna_ <varna_!~varna@> has quit IRC (Ping timeout: 240 seconds)
[15:40:10] *** varna <varna!~varna@> has joined #illumos
[15:50:03] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has joined #illumos
[16:02:49] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has joined #illumos
[16:10:09] *** ypankov <ypankov!~ypankov@> has quit IRC (Quit: leaving)
[16:19:59] *** jimklimov1 <jimklimov1!~jimklimov@ip-86-49-254-26.net.upcbroadband.cz> has joined #illumos
[16:39:23] *** tsoome_ <tsoome_!~tsoome@7726-3c68-c395-1497-2f80-4a40-07d0-2001.sta.estpak.ee> has joined #illumos
[16:43:56] <tsoome> rmustacc: ping:)
[16:45:05] <jbk> it's early on the west coast :)
[16:45:14] <tsoome> aye:)
[16:45:22] <tsoome> just did realize:)
[16:46:08] <tsoome> btw, before I will dig into, does this sound legit: ../../sun4u/cpu/spitfire.c:3260: error: '0' flag used with '%b' cmn_err format [-Wformat] ?
[16:46:28] <rmustacc> tsoome: pong
[16:46:56] <rmustacc> am11: Mmm. We should probably add that part of c11. I punted on it in the past. We probably can. uchar.h is a C11 thing.
[16:47:00] <tsoome> rmustacc: hope i did not woke you up:)
[16:47:24] <rmustacc> No, you're fine. If IRC could wake me up, that'd be really bad!
[16:47:37] <tsoome> rmustacc: do you think we actually can nuke those CCVERBOSE (assuming check ofc)?
[16:48:44] <rmustacc> tsoome: So, personally I think it'd be clearer if we did them as separate passes. Same with the lint cleanup. Because then you can just wsdiff it and make sure it's the same and then also verify that we still have the -Wall -Werror output.
[16:49:05] <rmustacc> But if we're going to do multiple passes, let's at least file additional bugs for those changes so it's clear why that stuff was removed.
[16:49:16] <rmustacc> Erm, not multiple passes, but multiple things in the same pass.
[16:49:37] <rmustacc> am11: Though if we do add that, there is still a gotcha that most illumos distros won't have support for that for a bit, like the other bit.
[16:50:38] <tsoome> mkay. well, I only did include because there is long list of the makefiles involved. actually it is even longer than just drivers as virtually every component has it:D
[16:53:33] <tsoome> So I guess it would be good to have some clenup, if it would be better to implement as separate update or have it complementing other change(s), I do not know which approach is better:)
[16:57:02] <rmustacc> So, my rule of thumb around this is kind of two fold. One end is testing and one end is ease of understanding why it was done.
[16:57:28] <rmustacc> So, often when I'm adding functionality, clenaing up compiler warnings can make sense because there is the ability to leverage shared testing.
[16:57:51] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has quit IRC (Ping timeout: 260 seconds)
[16:57:58] *** amrfrsh <amrfrsh!~Thunderbi@> has joined #illumos
[16:59:39] <tsoome> well, in this case the issue is that original build system setup (in terms of compiler options) is done for studio cc, then the options were translated to gcc, and incidentally we just happen to get the same setup from 2 diffrent studio options.
[17:05:24] <tsoome> of course, we can word this issue another way around — if we did drop studio, then we should stop translating studio flags and replace them:)
[17:07:01] <LeftWing> richlowe is certainly, as I recall, keen to have GCC flags directly in the Makefiles
[17:08:48] <andyf> LeftWing: it was actually spotted during the review https://illumos.topicbox.com/groups/developer/Tc6a99322a4c0effd-M794067f1e4f05b7db3d357d5
[17:09:39] <andyf> LeftWing: I suppose for new packages we could ask if it been installed on an OpenIndiana host at least
[17:09:52] <andyf> LeftWing: I'll look to see if we can use `pkglint` a bit more to spot these
[17:10:09] <andyf> tsoome: Ok, I'll pick it up - just saw you as the assignee
[17:10:38] <LeftWing> andyf: Yes I think we need to be more clear and explicit with testing notes
[17:10:38] <tsoome> ok, thanks:)
[17:11:04] <LeftWing> "I used onu on an OI system and verified that the driver installs and attaches" is much clearer than "I tested it" say
[17:11:05] <andyf> I only picked it up because one of the tests we do for a new omnios release is a `pkg install *`
[17:11:42] <rmustacc> tsoome: I'd probably focus on doing the initial clean up without redoing the flags here, just to keep it more straightforward.
[17:14:57] <tsoome> yea, redoing flags is big one:)
[17:15:19] <rmustacc> One can still do the CCVERBOSE clean up before the other bit is done if it makes sense.
[17:16:51] <toastersonerson1> LeftWing (IRC): Thanks for the tip with /etc/path_to_inst now the vagrant libvirt box works .)
[17:16:58] <LeftWing> toastersonerson1: Nice!
[17:17:02] <tsoome> yep, hence I was picking on it:)
[17:20:18] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has quit IRC (Remote host closed the connection)
[17:29:25] *** dlg_ <dlg_!~dlg@toy.eait.uq.edu.au> has joined #illumos
[17:30:26] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 256 seconds)
[17:50:16] <gitomat> [illumos-gate] 12579 parallelize crypto-test build -- Patrick Mooney <pmooney at pfmooney dot com>
[17:52:51] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has joined #illumos
[17:55:23] <ptribble> Someone asked about legacy actions in packaging - Tribblix parses these, if they exist, but only to mine descriptive text; in particular I ignore the old SVR4 name they might contain
[17:55:46] <LeftWing> That's good to know, ptribble, thanks
[17:55:48] <andyf> Thanks ptribble
[17:56:02] <LeftWing> You know I think I recall finding the script you use to make the package names once haha
[17:56:04] <andyf> I've pushed up a review to at least fix the mlxcx manifest so there is no duplicate
[17:56:07] <LeftWing> I was looking for the c-runtime thing
[17:57:20] <ptribble> Yes, I wrote the convert IPS FMRI to SVR4 name thing a *very* long time ago. I daren't change it now as I don't really want packages to randonly change names
[17:58:28] <LeftWing> Yeah I can imagine you would need a lot of unit testing to ensure things remain the same in the face of modifications
[18:01:33] <andyf> ptribble: so you would not care if a new package didn't have a legacy action at all?
[18:02:06] <andyf> or, in particular for this one, if the mlxcx one lost its broken one rather than me fixing it?
[18:08:10] *** man_u_ <man_u_!~manu@fob.gandi.net> has joined #illumos
[18:08:54] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has quit IRC (Ping timeout: 240 seconds)
[18:08:54] *** man_u_ is now known as man_u
[18:11:02] *** chrisbf <chrisbf!~quassel@host86-137-4-2.range86-137.btcentralplus.com> has joined #illumos
[18:11:03] <rmustacc> andyf: Um, we haven't added them to a lot.
[18:11:11] <rmustacc> So, i40e, the cpu sensors, the tests, etc.
[18:11:20] <rmustacc> So I think it's pretty safe.
[18:12:46] <ptribble> It's perfectly safe without
[18:13:17] <ptribble> And I would say generally that new packages shouldn't have them
[18:14:31] *** man_u_ <man_u_!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has joined #illumos
[18:15:19] *** man_u <man_u!~manu@fob.gandi.net> has quit IRC (Ping timeout: 260 seconds)
[18:15:19] *** man_u_ is now known as man_u
[18:21:13] <andyf> Thanks - I will adjust the patch
[18:21:39] *** chrisbf <chrisbf!~quassel@host86-137-4-2.range86-137.btcentralplus.com> has quit IRC (Quit: https://quassel-irc.org - Chat comfortably. Anywhere.)
[18:21:44] <andyf> also `licence cr_SUN` creeps in to packages due to copied manifests..
[18:30:00] <rmustacc> We should write a guide to writing manifest, otherwise it'll probably keep happening.
[18:34:16] <andyf> I did not know that IPS enforces uniqueness on legacy pkg, learnt something
[18:34:45] <rmustacc> There's a lot of things that IPS enforces that I wish it'd actually check at repo build time. :/
[18:35:04] <andyf> It is on my list, I will take a look once we get r151034 out the door!
[18:36:01] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has quit IRC (Quit: man_u)
[18:36:09] <andyf> and open an issue in a bit
[18:37:03] <rmustacc> I hit that in the past due to a bunch of symlink or similar issues.
[18:37:32] <andyf> We do a lot of checking for omnios userland packages - I'm sure some of it is applicable to gate
[18:37:45] *** ypankov <ypankov!~ypankov@> has joined #illumos
[18:38:01] <andyf> although I'm not sure what level of IPS commands I can assume are present on a build host
[18:38:39] <rmustacc> Well, right now we already require a good chunk of the tool set, regardless of the 'p' option.
[18:45:04] *** wl_ <wl_!~wl_@2605:6000:1b0c:46ad::87c> has quit IRC (Quit: Leaving)
[18:48:39] <jbk> one thing that might be nice is a 'make fix' (or whatever name sounds good) target in usr/src/pkg
[18:48:59] <jbk> given that we still use and require legacy formatted manifests
[18:49:22] <jbk> which requires using undocumented env vars with pkgfmt
[18:49:54] <jbk> it's always mildly obnoxious when messing with manifests that i can never get it 100% correct for pkgfmt
[18:50:05] <jbk> and have to git out the right syntax to make it fix it for me
[18:50:08] <jbk> then commit that
[18:50:43] <andyf> in a bldenv, `pkgfmt <file>.mf` should do it for you
[18:50:54] <jbk> when we could just semi-automate it (semi in that you still need to explicitly request, but without having to dig out all the pkg arcana)
[18:51:40] <andyf> we could move to v2 manifests too..
[18:51:50] <andyf> it is not a big difference, just sort order for a couple of elements
[18:58:18] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has joined #illumos
[19:00:23] <am11> rmustacc: just to re-reconfirm, answer to janvorli's question https://github.com/dotnet/runtime/pull/35173#discussion_r410863333 is yes?
[19:05:37] <am11> if you or someone from Illumos can directly comment on PR, that would be better. :)
[19:07:07] <rmustacc> I am doing so.
[19:18:17] <rmustacc> am11: I've left some notes on that for you.
[19:21:20] <am11> rmustacc: thanks a lot!
[19:23:01] <am11> i missed the last comment about C11 (ISO 19769) uchar.h header, is that a mandatory header to qualify full ISO C11 support or an optional one?
[19:26:12] <rmustacc> am11: Not sure if it matters or not.
[19:26:38] <rmustacc> It doesn't exist today so even if I go add it, it's not going to help for existing software.
[19:27:55] <rmustacc> Though I might look at adding it.
[19:30:14] *** jellydonut <jellydonut!~quassel@s91904423.blix.com> has quit IRC (Quit: jellydonut)
[19:30:41] *** jellydonut <jellydonut!~quassel@s91904423.blix.com> has joined #illumos
[19:44:30] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has quit IRC (Remote host closed the connection)
[20:01:23] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has joined #illumos
[20:05:35] <gitomat> [illumos-gate] 12526 Could combine misc_link_i386.c handlers -- John Levon <john.levon at joyent dot com>
[20:31:11] *** Qatz <Qatz!~db@2601:187:8400:5::c92> has quit IRC (Read error: Connection reset by peer)
[20:31:32] *** Qatz <Qatz!~db@2601:187:8400:5::c92> has joined #illumos
[21:08:47] *** ypankov_ <ypankov_!~ypankov@> has joined #illumos
[21:11:51] *** ypankov <ypankov!~ypankov@> has quit IRC (Ping timeout: 250 seconds)
[21:36:46] *** BOKALDO <BOKALDO!~BOKALDO@> has quit IRC (Quit: Leaving)
[21:42:44] *** ballew <ballew!sid244342@gateway/web/irccloud.com/x-dphdyhgymsvaqaxg> has quit IRC (Ping timeout: 265 seconds)
[21:42:54] *** ballew <ballew!sid244342@gateway/web/irccloud.com/x-tvtcvgxnzcuosmgz> has joined #illumos
[21:53:52] *** liv3010m <liv3010m!~liv3010m@77-72-245-190.fibertel.com.ar> has quit IRC (Ping timeout: 256 seconds)
[21:54:44] *** liv3010m <liv3010m!~liv3010m@77-72-245-190.fibertel.com.ar> has joined #illumos
[22:04:08] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Read error: Connection reset by peer)
[22:08:04] *** ypankov_ is now known as ypankov
[22:09:38] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has quit IRC (Remote host closed the connection)
[22:10:31] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[22:17:15] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has joined #illumos
[22:17:51] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has quit IRC (Max SendQ exceeded)
[22:21:17] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has joined #illumos
[22:43:18] <am11> rmustacc: it was a little bit tricky but that missing C11 issue was fixed. now we have PAL test results from SmartOS: 6 out of 710 failing. pr: https://github.com/dotnet/runtime/pull/35218.
[22:46:40] <am11> the `DIRECTIO_ON` failure is something other PR is adding https://github.com/dotnet/runtime/pull/35173/files#diff-5184456371fe15a62bf4f49cec4bb9efR682 , i have probably done something horrible there or maybe the tests needs adjusting, or maybe it is because i am running the tests in Global Zone... not sure at this point.
[22:54:54] <jbk> KungFuJesus: did you ever hear back from Saso? As I said the big holdup I had w/ using his code was the lack of copyrights -- if he can send me an email with the correct values to use, i think i'd be good
[23:00:29] <jbk> am11: I think you might just need an updated platform
[23:00:53] <jbk> directio() wasn't 'supported' on zfs
[23:00:56] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[23:01:11] <jbk> but really zfs already provides a lot of the same semantics
[23:01:32] <jbk> so there was a change now where it basically just always returns success when called on a zfs filesystem
[23:01:50] <jbk> that was in feb or so.. so if you're running something older than that, it'll return an error
[23:01:58] *** pcd <pcd!~pcd@openzfs/developer/pcd> has quit IRC (Remote host closed the connection)
[23:02:20] *** pcd <pcd!~pcd@openzfs/developer/pcd> has joined #illumos
[23:06:10] <am11> jbk: i build and executed those tests on a VM running `SmartOS 20190829T000927Z x86_64`. would it make sense i we try to support a little older versions?
[23:06:39] <am11> iow, any way to detect / workaround directio for older version?
[23:08:24] <am11> i can switch to the latest version of SmartOS (or some other?) i that is recommended. :)
[23:09:07] <jbk> well just that failure would be expected on older illumos distros without that change
[23:09:35] <jbk> so i guess it depends on how much the runtime is requiring/depending on that
[23:16:13] *** andy_js <andy_js!~andy@> has quit IRC (Quit: andy_js)
[23:20:05] <am11> jbk: ok, thanks. i will prepare a latest SmartOS VM (unfortunately there is no official vagrant box available.. that's why i was using 2019 and earlier 2018 boxes, with VirtualBox support).
[23:26:43] <sjorge> am11 interesting, so one could target stuff for ‘sunos’ that would work for both Oracle Solaris and illumos, but also target illumos for all variants or SmartOS for SmartOS specific things?
[23:31:18] <jbk> it would be useful at some point to separate the two
[23:31:36] <am11> sjorge: if there are binary compatible, then it should work. otherwise we can specialize / group them as needed. i am not fully aware of compatiblity story among SunOS derived distros.
[23:31:45] <jbk> e.g. if we're able to add the sysconf parameter (or such) like rmustacc has suggested, that's only going to be on illumos
[23:32:00] <jbk> we have no way to make oracle follow what we do
[23:32:01] <toastersonerson1> is smartos diverged so strongly from illumos already?
[23:32:14] <toastersonerson1> oh oracle yes them
[23:32:29] <jbk> err i thought it was solaris/illumos
[23:32:38] <jbk> smartos and illumos should be fairly close
[23:32:59] <toastersonerson1> it can distinguish between both
[23:33:08] <toastersonerson1> and smartos was talked about last :)
[23:33:12] <am11> basically the nuget packages for 3rd party as well as the runtime rely on RID to distinguish between the platforms (e.g. linux-musl-x64 works on both Alpine and Void Linux musl edition), but Android (bionic-libc based distro) can't use neither linux-x64 (glibc) nor musl-libc ones.
[23:33:15] <jbk> we merge in the latest illumos-gate (week)daily (not that that's a work, but on work days)
[23:33:31] <jbk> and a lot of the bits that aren't in illumos-joyent are just things we haven't had the time to upstream
[23:33:36] <jbk> and most of those shouldn't make a difference for this
[23:33:50] <jbk> unless the CLR is digging into stuff like OS packaging
[23:33:55] <jbk> (which i'd hope it's not)
[23:34:06] <sjorge> Maybe using OmniOS Next is a good target?
[23:34:25] <sjorge> The one andy is prepping now
[23:34:36] <sjorge> Its also the target for rust right?
[23:34:37] <jbk> generally as far as the baseos is concerned, things should largely be ABI compatible
[23:34:41] <am11> +1, will add OmniOS to my build matrix. :)
[23:35:46] <toastersonerson1> i think we should only have illumos as build target. not to make the clr people think that distributions are diverging.
[23:35:50] <sjorge> Since you seem to be doing most things in the gz on smartos, i think omnios (or openindiana) might be a better dev platfrom
[23:36:02] <sjorge> toastersonerson1++
[23:36:13] <toastersonerson1> we have vagrant boxes :)
[23:36:31] <toastersonerson1> https://app.vagrantup.com/Toasterson/boxes/openindiana-hipster
[23:36:45] <sjorge> Indeed
[23:37:23] <sjorge> Does that already have the above mentioned directio fix?
[23:37:40] <toastersonerson1> yes its from 5 days ago
[23:37:49] <toastersonerson1> 7
[23:40:30] <am11> great! so if we make an analogy of Fedora vs. Ubuntu, where former uses the (raw) Linux kernel un-patched, and latter has fork with all kinds of cleanups of old drivers etc. in place; is OpenIndiana more "pure" than SmartOS or are the the same when it comes to consuming Illumos base os?
[23:41:06] <toastersonerson1> openindiana uses upstream illumo-gate
[23:41:13] <sjorge> I’d say the same
[23:41:30] <sjorge> And openindiana is sort of the de facto reference distro
[23:42:08] <sjorge> Say a Debian if you will, while OmniOS would be closer to a centos, and SmartOS would be closer to proxmox
[23:42:13] <sjorge> But they are all illumos
[23:42:13] <toastersonerson1> It should be the actual refernce distro aswell. although all are equal these days due to the many patches which add function and the different roles.
[23:42:30] <am11> +1, I will switch to OI then. :)
[23:42:48] * sjorge hopes he didn’t offend any distro with that comparison
[23:43:17] <toastersonerson1> well comparing us to linux distros is almost impossible. we are too different.
[23:43:25] <sjorge> I think it might make your life a bit easier am11 for the dev work
[23:43:35] <sjorge> And that is a plus
[23:43:49] <toastersonerson1> for all intents and purposes all distros are one os.
[23:44:04] <sjorge> toastersonerson1: form sure, it was more of a general purpose/desktop, vs more server, vs hypervisor
[23:44:22] <am11> i hope the same thing for my comparison, no offense to Debian folks. i heard that for this very reason Torvald's uses CentOS/FreeBSD, but could be a rumor. :D
[23:44:29] <sjorge> Anyway, time to cache some Z’s
[23:44:45] <toastersonerson1> yep me too.
[23:45:05] <am11> btw, one of the arithmatic test failed: `atan2(0, -0) returned 0 when it should have returned 3.1415926535897931` from https://github.com/dotnet/runtime/blob/faebda75e202d810b00ce8176c79363006f62113/src/coreclr/src/pal/tests/palsuite/c_runtime/atan2/test1/test1.cpp#L92.
[23:45:06] <sjorge> Torvalds uses FreeBSD 🤔
[23:45:26] <am11> ah Fedora*
[23:45:28] <am11> lol
[23:47:08] <am11> * freedom is not free... unless it is FreeBSD, then it is free *
[23:48:14] <am11> (ops, didn't mean to bold that meme)
[23:53:03] *** ypankov_ <ypankov_!~ypankov@> has joined #illumos
[23:56:27] *** ypankov <ypankov!~ypankov@> has quit IRC (Ping timeout: 260 seconds)

   April 20, 2020  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | >