dotnet
LOGS
14:34:33 <Rhea> #startmeeting Fedora DotNet (2018-10-18)
14:34:33 <zodbot> Meeting started Thu Oct 18 14:34:33 2018 UTC.
14:34:33 <zodbot> This meeting is logged and archived in a public location.
14:34:33 <zodbot> The chair is Rhea. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:34:33 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:34:33 <zodbot> The meeting name has been set to 'fedora_dotnet_(2018-10-18)'
14:34:38 <Rhea> #meetingname dotnet
14:34:38 <zodbot> The meeting name has been set to 'dotnet'
14:34:40 <Rhea> #nick dotnet
14:34:52 <Rhea> #topic Packaging progress / Open Floor discussion
14:34:55 <tmds> .hello tmds
14:34:56 <zodbot> tmds: tmds 'Tom Deseyn' <tom.deseyn@gmail.com>
14:34:58 <aslice> .hello aslice
14:34:59 <zodbot> aslice: aslice 'Andrew Slice' <andrew.slice@redhat.com>
14:35:09 <omajid> .hello omajid
14:35:10 <zodbot> omajid: omajid 'Omair Majid' <omajid@redhat.com>
14:35:37 <Rhea> #chair omajid tmds aslice dseefeld crummel
14:35:37 <zodbot> Current chairs: Rhea aslice crummel dseefeld omajid tmds
14:36:21 <Rhea> So last time we were talking about CI
14:36:52 <Rhea> I guess that it's safe to say that CI with Rawhide is too much trouble to maintain and it wouldn't really be CI by definition but basically manual work...
14:38:00 <omajid> Rhea: can you elaborate more on the context of CI? if it's just building .net core, for example, historically rawhide has not been too painful
14:38:39 <omajid> thinking back, i dont think too many builds have failed where the failure was rawhide-specific
14:38:56 <Rhea> Well, we don't have an image, so I would have to maintain one if we wanted it... And I don't have resources to spare for that.
14:39:13 <omajid> ah. okay, so by CI you meant something in your qe framework. got it.
14:39:13 <Rhea> The problem is with maintaining the rawhide image itself, not with ci builds.
14:39:26 <omajid> i see. okay. please ignore what i csaid, then.
14:39:28 <omajid> said*
14:40:23 <Rhea> Every update would be manual, so at least once a month... and it doesn't matter what kind of infrastructure it would be.
14:40:53 <Rhea> Unless you want a ... `dnf upgrade` in every CI build (which i bet would break every build haha)
14:41:45 <crummel> yeah, that seems like it would cause problems
14:42:34 <Rhea> The next best thing to do is imho keeping it on the latest released version, i.e. right now that would be F29 already
14:44:02 <crummel> yeah, looking at our F28 Dockerfile it doesn't seem like we're doing anything crazy so we can probably keep on top of the latest released version pretty easily
14:46:33 <Rhea> When we're done with all the chaos we had this week I'd really like to look into CI for various things that were on my backlog for a while... so expect me bugging you omajid aslice tmds about that in near future :)
14:46:57 * omajid would be happy to help
14:47:09 <Rhea> crummel: do you have anything to bring up today, any issues or progress to talk about?
14:48:47 <crummel> I don't think we have much.  Davis merged the PR for 2.2 preview 3 so we'll be testing that out and hopefully have it tagged for you soon
14:49:46 <omajid> is this something we want to make avaiable to users (for playing around with, it's not stable/final)?
14:51:09 <omajid> i definitely dont want to put it in https://copr.fedorainfracloud.org/coprs/g/dotnet-sig/dotnet/ but if there's a need to, we could put it in some beta copr channel?
14:51:14 <Rhea> omajid: hmm... only as separate copr repo, I wouldn't want to mix it with stable to be honest.
14:51:19 <Rhea> Haha same thoughts
14:51:21 <omajid> agreed :)
14:51:58 <omajid> otherwise i can do builds in my not-meant-for-public-consumption copr repo for testing
14:52:07 <Rhea> I personally, as a user, am after stable things. However Fedora does have a lot of tinkers who like the latest bleeding edge so dunno..
14:54:11 <Rhea> omajid: so yeah up to you, I'm not against it being in... `@dotnet-sig/dotnet-preview` or -beta ?
14:54:31 <Rhea> If you want to build it regardless, and it's a question of where...
14:54:48 <omajid> okay, let me think about it a bit more
14:55:16 <Rhea> If we do go this way I'll include reference on the dev portal, etc...
14:56:46 <omajid> okay. i will get back to you about this.
14:57:37 <Rhea> Anyone got anything else to add, something to talk about? :)
14:58:09 * aslice shakes his head.
14:58:12 <crummel> not from our side
14:58:45 <omajid> there's one thing i want to bring up. not expencting an answer, but just to raise awareness
14:58:56 <Rhea> Bring it! O_O
14:58:57 <omajid> what's our stable native/unmanaged api?
14:59:38 <omajid> the context here is that some programs (looking at https://github.com/Samsung/netcoredbg for now) might want to link against coreclr
15:00:12 <omajid> and for that, it would be useful to know if coreclr has a public api/abi that they will keep stable (at least within a release)
15:00:48 <omajid> otherwise, for tools like this, everytime we update .net core, we would have to rebuild these programs to make sure they continue working
15:01:45 <omajid> just something to think about.
15:01:45 <crummel> ok, that makes sense.  I'm not sure what the story is there, I will bring this to Lee and the others and see what they say.
15:02:48 <Rhea> Kinda brings me to... I need to fire up slack.
15:02:59 <omajid> the other aspect to that quesiton is: if there is a stable public api, do we want to publish them at some well known paths? like /usr/include/coreclr.h or something?
15:05:22 <crummel> Okay, I'll bring that up too, maybe I should make a CoreCLR issue for it
15:08:27 <Rhea> omajid: thoughts on creating the issue right away?
15:08:36 <Rhea> And to continue discussion there
15:08:51 <omajid> sure, i can file an issue
15:09:36 <Rhea> Okay o.o
15:09:45 <Rhea> Anything else anyone?
15:10:28 <omajid> nothing else form me
15:10:55 <Rhea> #endmeeting