13:01:28 <petersen> #startmeeting Haskell SIG
13:01:28 <zodbot> Meeting started Tue Jun  9 13:01:28 2020 UTC.
13:01:28 <zodbot> This meeting is logged and archived in a public location.
13:01:28 <zodbot> The chair is petersen. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:28 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:01:28 <zodbot> The meeting name has been set to 'haskell_sig'
13:02:07 <tristanC> hello o/
13:02:27 <petersen> #meetingname haskell-sig
13:02:27 <zodbot> The meeting name has been set to 'haskell-sig'
13:02:34 <petersen> hey tristanC!
13:03:52 <petersen> I started working a little on f33 work
13:04:09 <petersen> We do have a few tickets https://pagure.io/haskell-sig/issues
13:05:22 <tristanC> oh nice, didn't knew about that space, i'll watch it
13:05:41 <petersen> I also fixed one longstanding issue in cabal-rpm when updating with subpackaging (yeah sigh subpackaging)
13:05:45 <petersen> tristanC: cool
13:06:32 <petersen> Well those are the first 4 tickets - only really started using it quite recently, but I thought it is good to track work
13:06:46 <tristanC> about unbundling, after  `cabal-rpm spec`, `fbrnch create-review`, `fbrnch request-repos`, i'm now waiting for https://pagure.io/releng/fedora-scm-requests/issue/25628 to complete time-manager
13:06:59 <petersen> Also one user complaining that cabal-install is too old, which is fair but not much we can do other than create a module perhaps
13:07:25 <petersen> tristanC: thanks okay
13:07:36 <tristanC> #link https://pagure.io/releng/fedora-scm-requests/issue/25628
13:07:43 <petersen> Usually repos are created quite promptly
13:08:21 <petersen> tristanC: perhaps we can coordinate on the push to rawhide cos it it is better to update the dependent packages too at the same time
13:09:00 <petersen> It is probably not worth backporting to older releases, at least immediately
13:09:24 <tristanC> petersen: yes, i'll get in touch before providing the standalone ghc-time-manager
13:09:41 <petersen> Okay sure thanks
13:09:58 <tristanC> i can actually prepare the PR to remove the bundle from the affected package
13:10:08 <petersen> I opened https://pagure.io/haskell-sig/issue/4 for all the current subpackaging
13:10:17 <petersen> tristanC: okay sure thanks
13:10:44 <tristanC> excellent, then i'll comment with all the PR associated with time-manager
13:11:08 <petersen> I think only two packages are affected: http2 and warp iirc
13:11:19 <petersen> Cool, thank you
13:12:07 <petersen> Oh and I updated Agda (and rebased Agda-stdlib) in Rawhide
13:12:33 <tristanC> i had a bad experience after enabling the 8.10 module when i had a few ghc-*-devel package from the non-module repository. not sure what's up with that
13:12:41 <petersen> Actually in Agda-2.6.1 the Edison deps disappear so that is good
13:12:54 <tristanC> also someone requested the availability of a more recent ghc outside of module here: https://lists.fedoraproject.org/archives/list/haskell@lists.fedoraproject.org/thread/4WCGJ2CUJEUHMF27TPODD2ZRL6PQNT24/
13:12:55 <petersen> tristanC: oh
13:14:15 <petersen> tristanC: I had missed that mail, thanks - I will try to respond
13:15:02 <petersen> Yeah people always think they want newer ghc... I dunno, I suppose it is somewhat reasonable but 8.6.5 works pretty well too...
13:15:24 <petersen> Stackage LTS 16 was just released btw (still 8.8.3)
13:16:10 <petersen> But in a way I feel modules are not a strong fit unless we actually use them to build stuff in Fedora itself
13:16:26 <petersen> Actually can parallel install very well - no problem at all
13:16:29 <petersen> oops
13:16:49 <petersen> Actually ghc can parallel install perfectly well
13:17:17 <petersen> Just need version suffices for binaries
13:17:45 <petersen> Long "we" more or less did this but then at bos' suggestion I dropped it
13:18:58 <tristanC> basically my issue can be reproduced like so: `podman run -it --rm fedora bash -c "dnf install -y ghc-time-manager-devel && dnf module enable ghc:8.10 && dnf update -y"`  that last dnf update will prints lots of problem
13:19:02 <tristanC> such as `Problem 1: package ghc-Cabal-3.2.0.0-93.module_f33+8393+d9235bfb.x86_64 requires libHSdeepseq-1.4.4.0-ghc8.10.1.so()(64bit), but none of the providers can be installed`
13:20:07 <tristanC> thus i worry those modules doesn't fit well with non-modules install
13:20:39 <petersen> Ah yeah you can mix and match ghc and main libraries alas
13:20:41 <petersen> cannot
13:21:05 <petersen> I don't think dnf can handle it more gracefully
13:21:13 <petersen> Unless I am missing something
13:21:41 <petersen> Basically you have to `dnf remove ghc-base` before install the module
13:21:49 <petersen> 'ing
13:23:16 <tristanC> yeah, then i'm happy with sticking to lts version from the main fedora repository
13:23:24 <petersen> Okay
13:24:06 <petersen> I often wish things were looser but yeah I feel the pain a lot
13:25:49 <petersen> There is one annoying misfeature in cabal-rpm with the recent doc files permissions fix
13:25:53 <tristanC> what about cabal-install, iirc this is not part of stackage lts, but could we get a 3.x version in f33 or is this going to cause to much risk?
13:26:06 <tristanC> iiuc*
13:26:56 <petersen> Well it really should be in my opinion... unforunately cabal-install-3.0 does not build with ghc-8.8.3 🤦‍♂️️
13:27:18 <petersen> I _hope_ I can get it to build somehow ;o) lol
13:27:45 <tristanC> oh i see, that's unfortunate :-)
13:28:13 <petersen> Yeah upstream basically assume you use whatever Cabal they tell you too, sigh
13:28:34 <tristanC> In other news, I also meant to mention this fedora-magazine article proposal about getting started with haskell:
13:28:38 <tristanC> #link https://teams.fedoraproject.org/project/asamalik-fedora-magazine/us/182
13:28:39 <petersen> I think stack-2.3 should be okay for LTS 15/16
13:29:26 <petersen> tristanC: yes thank you - I enjoyed the read - I think you did a nice job there
13:29:51 <tristanC> petersen: thanks a lot for the feedback!
13:30:05 <petersen> I will go through it again and see if I have any more feedback, you're welcome
13:30:24 <petersen> Ah I have to run to the Workstation meeting
13:30:37 <tristanC> alright, thank you!
13:30:46 <petersen> Thanks tristanC for the meeting
13:32:15 <petersen> #endmeeting