| 1 Jun 2021 |
 maralorn | Yes, you could. | 13:39:23 |
| maralorn | Allthough if your project has not Setup.hs you will need to make one yourself. | 13:39:49 |
| maralorn | * Although if your project has not Setup.hs you will need to make one yourself. | 13:40:02 |
 sterni (he/him) | generally we should advertise shellFor more, development environments and build environments are different things | 14:08:41 |
| maralorn | True | 14:09:18 |
 joe (he/him) | I really like developPackage | 14:10:08 |
| maralorn | I have no clue what the difference is … | 14:10:28 |
| maralorn | Why do we have both? | 14:10:35 |
| sterni (he/him) | developPackage does some callCabal2nix magic | 14:10:51 |
| sterni (he/him) | plus it can be both the derivation and the shell | 14:11:00 |
| maralorn | Interesting | 14:11:36 |
 pnotequalnp | How can I nix shell GHC 9? I haven't been able to figure out what path to use to get stuff out of haskellPackages. | 17:12:15 |
| maralorn | pnotequalnp: Note that we don‘t have a lot of support for ghc 9 yet. But you can use haskell.packages.ghc901.<package-name> for ghc901 packages. | 17:13:27 |
| maralorn | haskell.compiler.ghc901 should be the package with the correct compiler. | 17:13:48 |
| maralorn | * haskell.compiler.ghc901 should be the package with the correct compiler. | 17:14:06 |
| maralorn | * pnotequalnp: Note that we don‘t have a lot of support for ghc 9 yet. But you can use haskell.packages.ghc901.<package-name> for ghc901 packages. | 17:14:16 |
 pennae | how does one write a proper derivation for a project that contains both haskell and c++ binaries that want to be shipped together (in the same bin dir) | 21:47:23 |
| pennae | because symlinkJoin'ing two separate derivations doesn't feel like it's it | 21:47:37 |
| sterni (he/him) | why not | 21:47:45 |
| maralorn | do they need to be linked together in any way? | 21:47:59 |
| sterni (he/him) | I mean you can copy them if you prefer that | 21:48:15 |
| pennae | no, they just have to call each other | 21:48:18 |
| pennae | symlinkJoin works perfectly, it just feels like a hack. but we have no clue what we're doing, so that feeling doesn't count for much :) | 21:48:58 |
| maralorn | pennae: Doesn‘t really seem like a hack to me. | 21:49:31 |
| sterni (he/him) | well symlinking has the advantage that you can build both projects independently, so build time is halved | 21:49:35 |
| sterni (he/him) | like imagine you fix a typo in the C++ thing, then you have to wait for the haskell build as well if it's one drv | 21:49:57 |
| maralorn | pennae: But you are free to dangle the build of one of them into the derivation of the other one. | 21:50:00 |
| pennae | dangle into..? | 21:50:21 |
| pennae | building independently is nice, definitely | 21:51:02 |
| maralorn | In reply to @pennae:matrix.eno.space
dangle into..? I thought that would be a word like "fiddle" "hack" in this case I meant "insert". Anyway what I wanted to say you can certainly solve it differently but it would be a lot more work. | 21:54:14 |
