| 3 Oct 2024 |
 jasagredo | Notice this is different from rust features. There you can specify enabled features on your dependencies, and that will be transitive for every other package depending on yours. Here in Cabal there is no way to do that. | 17:27:51 |
 6d03 | i suppose to be able to do this, cabal would have to potentially instantiate the same package with different flag settings and it just can't. is that it? | 17:31:13 |
 Sean | You can add it to your cabal.project
| 18:02:09 |
| jasagredo | No, it doesn’t have to do with having the same package again with different flags. The flags of dependencies are simply not settable in cabal files, therefore not distributable, only local | 18:04:56 |
| Sean | I don't think it's a technical limitation, it's seen as the responsibility of the end developer
| 18:05:16 |
| jasagredo | If it was a problem of having the same package twice, then private dependencies would solve it (when that is finished and merged). But it is not | 18:05:49 |
 sm | In reply to @fendor:matrix.org
HLS CI encountered some weird issues on MacOs which resolved itself a couple of days ago... Does somebody know what might have caused this, e.g., was it a GHA issue? https://github.com/haskell/haskell-language-server/actions/runs/10760272709/job/29838001624?pr=4399 I saw that too, I forget how it went away.Maybe I cleared caches ? | 21:48:40 |
| sm | In reply to @fendor:matrix.org
HLS CI encountered some weird issues on MacOs which resolved itself a couple of days ago... Does somebody know what might have caused this, e.g., was it a GHA issue? https://github.com/haskell/haskell-language-server/actions/runs/10760272709/job/29838001624?pr=4399 * | 21:49:18 |
| sm | In reply to @6d03:matrix.org
i suppose to be able to do this, cabal would have to potentially instantiate the same package with different flag settings and it just can't. is that it? There may be other reasons it’s this way, but I imagine one is that we already were struggling to handle deps sanely with just packages and components, adding random flag flavours on top would have been chaos. And it’s not necessary, better to enforce have some discipline on ourselves | 21:55:18 |
| sm | * | 21:55:28 |
| sm | In reply to @fendor:matrix.org
HLS CI encountered some weird issues on MacOs which resolved itself a couple of days ago... Does somebody know what might have caused this, e.g., was it a GHA issue? https://github.com/haskell/haskell-language-server/actions/runs/10760272709/job/29838001624?pr=4399 * | 21:56:55 |
| 4 Oct 2024 |
|  Kyle Butt changed their profile picture. | 18:15:07 |
| 13 Oct 2024 |
|  hellwolf joined the room. | 18:07:26 |
|  @blaisepabon:matrix.org left the room. | 23:43:31 |
| 16 Oct 2024 |
|  Selayar Laut joined the room. | 06:23:09 |
| 21 Oct 2024 |
|  amesgen joined the room. | 15:53:01 |
|  (artur 'manuel) joined the room. | 19:57:10 |
| (artur 'manuel) changed their display name from (lambda (f l) (format nil "~a ~a")) "Artur" "Manuel" to (artur 'manuel). | 20:02:49 |
| 23 Oct 2024 |
|  dyniec changed their profile picture. | 12:10:52 |
| 26 Oct 2024 |
|  @ivarsfg:matrix.org left the room. | 00:55:18 |
| 27 Oct 2024 |
|  ~hedy joined the room. | 09:22:27 |
| 1 Nov 2024 |
|  @axiopaladin:matrix.org left the room. | 02:14:08 |
 Nadia | How good is git://github.com/elaforge/fix-imports.git ? | 16:42:06 |
 Hécate | never used it | 16:44:39 |
| 3 Nov 2024 |
 mouse009 | Dyes anybody know why GHC-9.8.3 is not in `ghcup`? | 23:57:56 |
| 4 Nov 2024 |
 Artem | @mouse009:matrix.org: apparently, Julian is moving to another country https://github.com/haskell/ghcup-metadata/issues/255 and ghcup is still pretty much one-man shop | 00:34:56 |
| mouse009 | I see, thanks.
A pity that such a universally used and popular tool remains a one-man shop. Any idea if Julian and/or the community are planning to improve this situation? | 02:29:42 |
 chreekat | The Haskell Foundation sponsors someone to help with ghcup. There are regular checkups, too. But keep in mind that ghcup is not interested in making the latest and greatest available as quickly as possible. "E.g. it frequently happens that upstream developers want a new release to be ‘recommended’ (that is: installed by default if you run GHCup for the first time). However, experience shows that it’s usually better to wait." https://discourse.haskell.org/t/ghcup-is-not-an-installer/8102 | 07:08:32 |
| Hécate | In reply to @mouse009:matrix.org
I see, thanks.
A pity that such a universally used and popular tool remains a one-man shop. Any idea if Julian and/or the community are planning to improve this situation? yes there are plans, can't share them in much detail right now because funding might still prove problematic, but rest assured there are dark and occult forces (me) working to improve it all | 08:50:47 |
| Hécate | mouse009: Just so you understand that's it's not arbitrary, there's work to be done on the GHC side to make sure that the binary distributions can be reliably tested by distributors (ghcup, linux distros, etc). And sometimes, there are bugs that slip through the cracks, but then it falls onto Julian to correct them all https://github.com/haskell/ghcup-metadata/pull/253#issuecomment-2424949059 | 08:58:31 |
