Logs: freenode/#haskell
| 2021-04-03 16:43:54 | <sclv> | I don't agree we should redefine what hackage is, I do think we should finally get support for overlays and better CI reporting across hackage |
| 2021-04-03 16:44:17 | <sclv> | I.e. packages need some Authoritative Place to live (all packages) |
| 2021-04-03 16:44:18 | <maerwald> | better CI reporting across hackage IS redefining it |
| 2021-04-03 16:44:22 | <Logio> | maerwald: that's a disincentive to rely on hackage in the first place |
| 2021-04-03 16:44:37 | <maerwald> | Logio: yes, it was just an example |
| 2021-04-03 16:44:37 | <sclv> | but providing good subsets and ensuring them should be easier to layer on top instead of needing only third party solutions |
| 2021-04-03 16:45:23 | <sclv> | ok, so this proposal is just "formalize and define pre-release process more"? that seems fine |
| 2021-04-03 16:46:22 | <Uniaika> | sclv: yes, it's not a technical change to the produced artifacts |
| 2021-04-03 16:46:40 | <sclv> | ok and there is also a pre-release process that currently exists. |
| 2021-04-03 16:46:45 | <sclv> | including head.hackage -- do you know about that? |
| 2021-04-03 16:46:53 | <sclv> | https://ghc.gitlab.haskell.org/head.hackage/ |
| 2021-04-03 16:47:23 | <Uniaika> | I know about it indeed |
| 2021-04-03 16:47:37 | <Uniaika> | again, it's not a technical proposal |
| 2021-04-03 16:47:42 | <Logio> | for me the whole point of having a consistent distribution from the language side seems a bit odd in general, if we're considering end users |
| 2021-04-03 16:47:49 | <Logio> | speaking as a gentoo user |
| 2021-04-03 16:48:16 | <maerwald> | gentoo is 80% policies, workflows and 20% patches... compare that with haskell ecosystem |
| 2021-04-03 16:48:32 | <sclv> | Uniaika: I get its not technical. I'm asking that even as a process proposal it would be good to specify how the process differs from the process now, where there is typically already an extended pre-release that various systems make available. I.e. if this proposal is accepted, what "action items" would have to be taken by what people to implement it? |
| 2021-04-03 16:48:39 | <Logio> | the end-user packaging is always going to be dictated by distro package managers with different constraints than the language ecosystem |
| 2021-04-03 16:48:39 | <Uniaika> | Logio: speaking as a Funtoo user, I can assure you that consistency is the key to a pleasing experience ;) |
| 2021-04-03 16:48:47 | <Uniaika> | ask the Arch users how their migrations go |
| 2021-04-03 16:49:04 | → | wroathe joins (~wroathe@c-68-54-25-135.hsd1.mn.comcast.net) |
| 2021-04-03 16:49:38 | <Uniaika> | sclv: wait, before we continue, did you read the proposal? |
| 2021-04-03 16:49:47 | × | zaquest quits (~notzaques@5.128.210.178) (Read error: Connection reset by peer) |
| 2021-04-03 16:49:47 | <sclv> | yes |
| 2021-04-03 16:50:13 | <Logio> | Uniaika: yes, but it feels to me that consistency in a language ecosystem is just duplication (or waste) of work, since the effort needs to be taken anyway at the distro level |
| 2021-04-03 16:50:28 | → | zaquest joins (~notzaques@5.128.210.178) |
| 2021-04-03 16:50:35 | <Uniaika> | sclv: what do the "Proposed change specification" and "operating costs" sections fail to answer? |
| 2021-04-03 16:50:45 | <Uniaika> | let's focus on what is lacking from what I wrote |
| 2021-04-03 16:50:55 | <sclv> | the proposed change section looks exactly like what already exists! |
| 2021-04-03 16:51:11 | <sclv> | the pre-releases are published and put on various systems, and people are encouraged to use them and update their packages |
| 2021-04-03 16:51:18 | <Uniaika> | sclv: we do not have any kind of efforts to actively prepare the ecosystem for the transition |
| 2021-04-03 16:51:18 | <sclv> | the announcement email just has slightly different wording |
| 2021-04-03 16:51:22 | <Uniaika> | that is a fact |
| 2021-04-03 16:51:24 | <geekosaur> | Logio, distros package things for their needs, not those of developers. I don't think anyone but maybe C/C++ programmers relies on a distro for their language or packages |
| 2021-04-03 16:51:32 | <Uniaika> | we do not mobilise any kind of volunteer teams |
| 2021-04-03 16:51:50 | <geekosaur> | certainly all of perl, pythin, and js will tell you right off not to use a distro package for either |
| 2021-04-03 16:51:56 | <Uniaika> | sclv: however I worded this badly, I'll correct |
| 2021-04-03 16:51:57 | <Uniaika> | thanks |
| 2021-04-03 16:52:02 | <sclv> | the proposal does not in the proposed section talk about mobilising a volunteer team |
| 2021-04-03 16:52:12 | <Logio> | geekosaur: yeah, but that's my point; you don't need to stall releases to guard end users from bad experiences, that's the distro's job |
| 2021-04-03 16:52:24 | <sclv> | a plan to have such a team would be good -- its just not there -- I'm not being negative, I'm trying to make sure we can pin down exactly what is different than before :-) |
| 2021-04-03 16:52:39 | <Uniaika> | Logio: that's called having the package management strategy of C and C++, and it has failed |
| 2021-04-03 16:53:20 | × | shalokshalom quits (~quassel@2a02:1748:dd5e:7f60:cf49:8384:7c93:3106) (Remote host closed the connection) |
| 2021-04-03 16:53:21 | <Uniaika> | sclv: I know you're not negative, and that is why I constantly ask you to be more specific, so that I can better address your remarks :) |
| 2021-04-03 16:53:28 | × | petersen quits (~petersen@redhat/juhp) (Quit: petersen) |
| 2021-04-03 16:53:49 | <sclv> | Uniaika: one change that would be welcome, but I think would accomplish little is that the announcements (such as https://discourse.haskell.org/t/ghc-9-2-1-alpha1-now-available/2286) don't tend to have language specifically encouraging maintainers to test the pre-release against their libs and ensure compatibility, nor do they point to easy places to install the pre-release from |
| 2021-04-03 16:54:04 | → | shalokshalom joins (~quassel@2a02:1748:dd5e:7f60:cf49:8384:7c93:3106) |
| 2021-04-03 16:54:08 | → | petersen joins (~petersen@redhat/juhp) |
| 2021-04-03 16:54:17 | <maerwald> | geekosaur: python can be reasonably well packaged on distro level, despite much higher overall complexity than C |
| 2021-04-03 16:54:22 | <sclv> | note that we probably would need a second announcement for this, since the first announcement is what notifies the various install tools (ghcup etc) that they can provide the release at all |
| 2021-04-03 16:54:22 | <maerwald> | haskell not |
| 2021-04-03 16:54:34 | × | gehmehgeh quits (~ircuser1@gateway/tor-sasl/gehmehgeh) (Quit: Leaving) |
| 2021-04-03 16:55:22 | <Uniaika> | sclv: yes, good! |
| 2021-04-03 16:55:24 | <Uniaika> | thank you |
| 2021-04-03 16:55:32 | <sclv> | so its a "two phase" pre-release plan with ghchq doing what it normally does, and then the "release compat strike-force" making sure a good pre-release is avail in the Usual Ways, following which they announce "hey everyone, its easy to test your lib against this pre-release and head.hackage, here's how, please go for it" |
| 2021-04-03 16:55:55 | <Uniaika> | yuup' |
| 2021-04-03 16:55:56 | × | ericsagn1 quits (~ericsagne@2405:6580:0:5100:7445:6b92:4b01:cfc6) (Ping timeout: 258 seconds) |
| 2021-04-03 16:56:02 | <sclv> | (and arguably the alpha is too early for this anyway) |
| 2021-04-03 16:56:05 | <Uniaika> | thanks for the wording |
| 2021-04-03 16:56:13 | <Uniaika> | yeah that's why I'm talking about release candidates and stuff |
| 2021-04-03 16:56:28 | <sclv> | :thumbsup: |
| 2021-04-03 16:56:37 | × | luke quits (~luke@bitnomial/staff/luke) (Quit: sleep) |
| 2021-04-03 16:57:06 | <sclv> | I do worry about emailing every maintainer. |
| 2021-04-03 16:57:33 | <sclv> | if someone wants to be an active maintainer they'll surely follow at least some list where its announced, and if they're not, bugging them won't help much |
| 2021-04-03 16:58:04 | × | __minoru__shirae quits (~shiraeesh@77.94.25.131) (Ping timeout: 268 seconds) |
| 2021-04-03 16:58:25 | <sclv> | what the strikeforce needs is a CI tool to test a large-enough subset of Current Hackage (using maybe the matrix tools and head.hackage or the like) that they can figure out what the big roadblocks are |
| 2021-04-03 16:58:34 | <sclv> | and some repo or place where they can organize and coordinate |
| 2021-04-03 16:59:27 | <sclv> | (trustees used to do a lot more of this, but people burnt out on it, especially due to hostile maintainer responses -- having the strikeforce be community volunteers Without Special Ops Rights as distinct from trustees could help with this) |
| 2021-04-03 17:00:06 | × | frozenErebus quits (~frozenEre@37.231.244.249) (Ping timeout: 240 seconds) |
| 2021-04-03 17:00:41 | × | stree quits (~stree@68.36.8.116) (Ping timeout: 240 seconds) |
| 2021-04-03 17:00:48 | <Uniaika> | 👍 |
| 2021-04-03 17:01:02 | → | merijn joins (~merijn@83-160-49-249.ip.xs4all.nl) |
| 2021-04-03 17:01:22 | <sclv> | i.e. there's a trustees-maintained cookbook for fixing packages to be compat with new ghc releases, it could be taken over by a new team, or kept there but with more contributors or etc: https://github.com/haskell-infra/hackage-trustees/blob/master/cookbook.md |
| 2021-04-03 17:01:23 | → | dyeplexer joins (~lol@unaffiliated/terpin) |
| 2021-04-03 17:04:08 | <Uniaika> | sclv: I updated the proposal, tell what you think: https://github.com/Kleidukos/ghc-proposals/blob/patch-1/proposals/0000-ghc-maintainer-preview.md |
| 2021-04-03 17:06:59 | <sclv> | a fw things -- I would mention head.hackage and the cookbook as resources the team could use, and also note that the team should help coordinate making ghc available through ghcup, the ppa, etc. And also note that the team may also seek to implement more CI infrastructure to ease the task of finding the incompat packages |
| 2021-04-03 17:07:16 | <sclv> | but its definitely clearer |
| 2021-04-03 17:07:46 | → | ericsagn1 joins (~ericsagne@2405:6580:0:5100:1b1e:6b9b:245f:d5e2) |
| 2021-04-03 17:07:54 | <Uniaika> | awesome, thanks |
| 2021-04-03 17:08:01 | <sclv> | you may also want to note that rcs already exist but don't have a deliniated length of time, nor is there a coordinated effor to ensure their widespread availablility |
| 2021-04-03 17:08:37 | <sclv> | so if i want to use a pre-release there's no One Place I can look now to find out "how do i install it most easily given my setup" |
| 2021-04-03 17:09:13 | <sclv> | (back in the day in the day we used to have this via the haskell platform being the One True Method and it packaging up pre-releases as well. this led to burnout too... there's a pattern here, lol) |
| 2021-04-03 17:09:33 | <maerwald> | I must also say I never received notification of imminent releases, unless I actively read #ghc, so ghcup always has a delay of a day or two. |
| 2021-04-03 17:09:40 | → | __minoru__shirae joins (~shiraeesh@77.94.25.131) |
| 2021-04-03 17:10:28 | <Uniaika> | yes, maerwald's suffering must end at once |
| 2021-04-03 17:10:39 | <maerwald> | lol |
| 2021-04-03 17:11:10 | <geekosaur> | the ghc wiki has a status page for each release that tracks these things |
| 2021-04-03 17:11:29 | <geekosaur> | e.g. expected rc and release dates |
| 2021-04-03 17:11:51 | <maerwald> | are you saying these are accurate? :D |
| 2021-04-03 17:12:41 | → | Sheilong joins (uid293653@gateway/web/irccloud.com/x-wqegmyveynqiexyu) |
| 2021-04-03 17:13:44 | <geekosaur> | they're generally updated as things slip |
| 2021-04-03 17:13:45 | <Uniaika> | < sclv> so if i want to use a pre-release there's no One Place I can look now to find out "how do i install it most easily given my setup" // Yep I address this problem by saying "we make it available in ghcup and the ubuntu PPA" |
| 2021-04-03 17:14:12 | → | stree joins (~stree@68.36.8.116) |
| 2021-04-03 17:14:18 | <maerwald> | geekosaur: that doesn't work |
| 2021-04-03 17:14:50 | <maerwald> | geekosaur: what works is notifying packagers etc of an imminent release, so when the announcement hits, some works has already been done and it can be shipped quicker. |
| 2021-04-03 17:16:45 | <maerwald> | as in: longer period between uploading releases and bindists and actual announcement |
| 2021-04-03 17:17:14 | <maerwald> | then things can be more in lock-step |
| 2021-04-03 17:18:18 | <sclv> | Uniaika: sure, making it available there helps, but also the announcement email (maintainer preview one, not ghchq one) saying "hey everyone, here's where you can get it from!" is good |
| 2021-04-03 17:18:29 | <sclv> | people need to be reminded of things constantly, or they forget |
| 2021-04-03 17:19:04 | → | codygman__ joins (~user@47.186.207.161) |
All times are in UTC.