Logs: freenode/#haskell
| 2021-04-03 16:01:10 | <Uniaika> | so that we make it available for people who need time before we ship it to users |
| 2021-04-03 16:01:17 | <tomsmeding> | which would be for the ecosystem-central but non-boot libraries, I guess |
| 2021-04-03 16:01:42 | <Uniaika> | yup' |
| 2021-04-03 16:01:55 | <Uniaika> | those libraries whose development doesn't follow closely the GHC development |
| 2021-04-03 16:02:01 | → | luke_ joins (~luke@bitnomial/staff/luke) |
| 2021-04-03 16:02:31 | luke_ | is now known as Luke |
| 2021-04-03 16:02:39 | Luke | is now known as luke |
| 2021-04-03 16:03:24 | <tomsmeding> | IIRC the ghc devs already have some trouble once in a while with busy maintainers delaying a new release of boot libs |
| 2021-04-03 16:03:56 | <tomsmeding> | but I guess this would be more "strongly advised" instead of "required" for the release |
| 2021-04-03 16:04:21 | <maerwald> | tomsmeding: there's also "prerelease" in ghcup |
| 2021-04-03 16:04:25 | <Uniaika> | tomsmeding: yeah the ghc dev team has solidified the control and maintenance of ghc boot libs |
| 2021-04-03 16:04:42 | <tomsmeding> | maerwald: very good point |
| 2021-04-03 16:04:45 | <Uniaika> | yup I could totally see ghcup have this Maintainer Preview in "pre-release" |
| 2021-04-03 16:05:12 | <Uniaika> | but the proposal is not about dictating how ghcup should bundle it, it's more of a schedule + communication thing |
| 2021-04-03 16:05:17 | → | kiweun joins (~kiweun@2607:fea8:2a62:9600:28af:d168:c6d2:ab82) |
| 2021-04-03 16:05:34 | <Uniaika> | and if you agree or disagree I'd like to ask you to reply to the Proposal thread 🙇 |
| 2021-04-03 16:05:44 | <tomsmeding> | I'm currently more of a casual/academic user of haskell, so probably not the right person to ask regarding this proposal :) |
| 2021-04-03 16:05:59 | <mpickering> | I am not sure what problem this solves |
| 2021-04-03 16:07:50 | → | carlomagno joins (~cararell@148.87.23.4) |
| 2021-04-03 16:09:16 | <Uniaika> | mpickering: it's literally in the "Motivation" paragraph… |
| 2021-04-03 16:09:56 | × | kiweun quits (~kiweun@2607:fea8:2a62:9600:28af:d168:c6d2:ab82) (Ping timeout: 258 seconds) |
| 2021-04-03 16:11:26 | → | kritzefitz joins (~kritzefit@212.86.56.80) |
| 2021-04-03 16:13:48 | → | mkDoku joins (~TheMule@aftr-37-201-195-134.unity-media.net) |
| 2021-04-03 16:17:02 | → | geekosaur joins (82650c7a@130.101.12.122) |
| 2021-04-03 16:18:06 | × | petersen quits (~petersen@redhat/juhp) (Quit: petersen) |
| 2021-04-03 16:18:08 | → | geowiesnot joins (~user@87-89-181-157.abo.bbox.fr) |
| 2021-04-03 16:18:23 | × | cole-h quits (~cole-h@c-73-48-197-220.hsd1.ca.comcast.net) (Ping timeout: 265 seconds) |
| 2021-04-03 16:18:28 | <geekosaur> | let's put it this way, we're already in 9.2 prerelease and there are ecosystem libs not yet updated for 9.0.1 |
| 2021-04-03 16:18:44 | → | petersen joins (~petersen@redhat/juhp) |
| 2021-04-03 16:20:14 | <Uniaika> | geekosaur: that is indeed a fact |
| 2021-04-03 16:21:48 | → | tpefreedom joins (~tpefreedo@98.125.189.230) |
| 2021-04-03 16:22:46 | → | mouseghost joins (~draco@87-206-9-185.dynamic.chello.pl) |
| 2021-04-03 16:22:46 | × | mouseghost quits (~draco@87-206-9-185.dynamic.chello.pl) (Changing host) |
| 2021-04-03 16:22:46 | → | mouseghost joins (~draco@wikipedia/desperek) |
| 2021-04-03 16:24:00 | × | son0p quits (~son0p@181.136.122.143) (Quit: Lost terminal) |
| 2021-04-03 16:25:30 | <Logio> | who is the "final user" that needs to wait for things to be perfect, and how much are they paying package maintainers to make it so? |
| 2021-04-03 16:25:47 | <Uniaika> | Logio: commercial adopters, end-users who develop applications? |
| 2021-04-03 16:26:33 | <Logio> | why would they want to wait to get the newer compiler, compared to other users? |
| 2021-04-03 16:26:48 | → | todda7 joins (~torstein@athedsl-09242.home.otenet.gr) |
| 2021-04-03 16:26:55 | <Uniaika> | IT'S LITERALLY WRITTEN IN THE PROPOSAL |
| 2021-04-03 16:27:00 | <Uniaika> | READ IT |
| 2021-04-03 16:27:15 | <Uniaika> | or pay me 50€ to spare you the burden of clicking two links. |
| 2021-04-03 16:27:33 | <Uniaika> | because the ecosystem takes too much time to transition |
| 2021-04-03 16:27:43 | <Uniaika> | because nobody has the energy to do things |
| 2021-04-03 16:27:44 | <Logio> | I did read it, but I feel like it assumes a lot of things that I don't take as fact |
| 2021-04-03 16:27:49 | <Uniaika> | Logio: such as? |
| 2021-04-03 16:27:52 | → | ezrakilty joins (~ezrakilty@97-126-95-37.tukw.qwest.net) |
| 2021-04-03 16:28:08 | <maerwald> | I can put it a bit more controversial: if hackage/haskell was a distro, it would be considered dysfunctional. But this is also a question of scope. |
| 2021-04-03 16:28:27 | <Logio> | like the fact that having a waiting time would improve things in a meaningful way |
| 2021-04-03 16:28:37 | <Logio> | also what maerwald said ^ |
| 2021-04-03 16:28:45 | <Uniaika> | maerwald: it's not controversial at all, as an ecosystem/language pair, Hackage/Haskell is widely dysfunctional |
| 2021-04-03 16:28:56 | <Uniaika> | that's a fact |
| 2021-04-03 16:29:25 | <Uniaika> | Logio: it's a "waiting time" only from an end-user point of view. It's still release management but for the ecosystem |
| 2021-04-03 16:30:49 | → | Tario joins (~Tario@201.192.165.173) |
| 2021-04-03 16:31:32 | <Logio> | I feel like that's the same as saying a fish is a dysfunctional bike rider |
| 2021-04-03 16:31:45 | × | paddymahoney quits (~paddymaho@cpe9050ca207f83-cm9050ca207f80.cpe.net.cable.rogers.com) (Remote host closed the connection) |
| 2021-04-03 16:32:20 | → | tzh joins (~tzh@c-24-21-73-154.hsd1.or.comcast.net) |
| 2021-04-03 16:32:22 | <Uniaika> | if you see as much relationship between a language and official library ecosystem as between a fish and a boat, this indeed explains quite a lot your point of view. |
| 2021-04-03 16:32:24 | × | ezrakilty quits (~ezrakilty@97-126-95-37.tukw.qwest.net) (Ping timeout: 265 seconds) |
| 2021-04-03 16:32:31 | <Logio> | there's a compiler and there's things that use a compiler, having a fixed working set composed of them is an independent proposition |
| 2021-04-03 16:32:36 | <Uniaika> | s/boat/bike/ |
| 2021-04-03 16:32:54 | × | jamm_ quits (~jamm@unaffiliated/jamm) (Remote host closed the connection) |
| 2021-04-03 16:33:08 | <Uniaika> | Logio: yet we have boot libs in GHC |
| 2021-04-03 16:33:14 | <Logio> | I mean, the question is what you consider "official" |
| 2021-04-03 16:33:31 | <Uniaika> | hackage |
| 2021-04-03 16:34:10 | × | conal_ quits (~conal@198.8.81.74) (Quit: Computer has gone to sleep.) |
| 2021-04-03 16:34:16 | <Uniaika> | it's hosted by haskell.org, the code is under the haskell namespace on GitHub, it's whence Cabal fetches dependencies by default |
| 2021-04-03 16:35:05 | → | ddellacosta joins (ddellacost@gateway/vpn/mullvad/ddellacosta) |
| 2021-04-03 16:35:22 | <lyxia> | Well it seems true that currently when a new GHC version comes out, people are excited but most can't use it right away because libraries have yet to catch up, that's a shame, one might want to fix that, and that's what this proposal does. |
| 2021-04-03 16:36:13 | <tomsmeding> | Uniaika: except nigh everybody can push to hackage? |
| 2021-04-03 16:36:41 | → | alx741 joins (~alx741@181.196.68.6) |
| 2021-04-03 16:36:55 | <Uniaika> | tomsmeding: Hackage trustees already curate packages to help with migrations |
| 2021-04-03 16:37:04 | <tomsmeding> | I think it's unrealistic to expect that ~the whole of hackage updates to a new ghc when a new one comes out |
| 2021-04-03 16:37:15 | <tomsmeding> | but it might be realistic to reduce that to popular, core libraries |
| 2021-04-03 16:37:33 | <Uniaika> | so there are numerous precedents of "official" Hackage trustees taking on their shoulders the burden of having everything hold together |
| 2021-04-03 16:37:40 | <fendor> | I don't think any waiting time would have made the transition from 8.10 to 9.0 any easier. The required changes are huge, e.g. even a couple of months after the release, HLS does not support it (iirc), and they have a bunch of very motivated (and capable) developers such as jneira. |
| 2021-04-03 16:37:44 | <Uniaika> | tomsmeding: it's unrealistic indeed, I shall put that in the proposition |
| 2021-04-03 16:37:58 | <maerwald> | I think the solution is much more radical and there's too little excitement in that direction: redefining what hackage is. |
| 2021-04-03 16:38:17 | <Uniaika> | maerwald: I didn't want to go down that road right away but indeed… |
| 2021-04-03 16:38:21 | <maerwald> | without excitement, you can't bring about change, even if you have the right ideas |
| 2021-04-03 16:38:39 | → | conal joins (~conal@64.71.133.70) |
| 2021-04-03 16:38:39 | × | conal quits (~conal@64.71.133.70) (Client Quit) |
| 2021-04-03 16:38:55 | <Uniaika> | fendor: I'm not dissing the maintainers of HLS, but not every library had such an enormous amount of work ahead of them |
| 2021-04-03 16:38:56 | → | __minoru__shirae joins (~shiraeesh@77.94.25.131) |
| 2021-04-03 16:39:03 | <maerwald> | ecosystem works for most ppl as is: ppl pin package versions via stack and nix or freeze files... almost no one does "rolling" anymore |
| 2021-04-03 16:39:20 | <Logio> | Uniaika: It does feel like you want hackage to be stackage |
| 2021-04-03 16:39:34 | <Uniaika> | Logio: it is not the case |
| 2021-04-03 16:39:40 | <Uniaika> | that's maerwald's idea :P |
| 2021-04-03 16:39:51 | → | conal joins (~conal@64.71.133.70) |
| 2021-04-03 16:40:13 | <fendor> | Uniaika, I am rather arguing that this particular transition is really a lot of work, one where I believe no amount of prep time would have made a difference |
| 2021-04-03 16:40:24 | → | maroloccio joins (~marolocci@pousada3ja.mma.com.br) |
| 2021-04-03 16:41:15 | <Uniaika> | fendor: and I agree with you |
| 2021-04-03 16:42:11 | <maerwald> | you have to create incentives for ppl to put in extra work to be compatible with newer GHCs... your package being dropped from hackage would be such an incentive |
| 2021-04-03 16:42:25 | <maerwald> | but probably too radical |
| 2021-04-03 16:42:34 | <Logio> | I get the idea of having release schedules aligned between developers in a single organization, I just don't see how that relates to hackage |
| 2021-04-03 16:42:51 | <sclv> | i'm actually confused about how this "maintainer preview" is substantively different from the pre-releases that now exist |
| 2021-04-03 16:43:02 | <sclv> | I think the proposal would need to explain the difference for anyone to understand |
| 2021-04-03 16:43:22 | <Logio> | maerwald: I don't get that kind of thinking, why should someones package be removed from hackage if it still works with some version of GHC? |
| 2021-04-03 16:43:33 | <maerwald> | pre-releases don't follow any process, they're done ad-hoc |
| 2021-04-03 16:43:48 | <maerwald> | Logio: to create incentives, as said above |
All times are in UTC.