 | sbt | 2088 Members |
| Scala Build Tool | 5 Servers |
| sbt | 2088 Members |
| Scala Build Tool | 5 Servers |
| Sender | Message | Time |
|---|---|---|
| 11 May 2024 | ||
 matthiasberndt | Yeah well, some think mill was a mistake. I think sbt was a mistake. | 01:20:16 |
 anemish | I agree. I told that sbt, CanBuildFrom, slow compiler and breaking changes basically killed any chance for Scala to become mainstream. But I don’t see how Scala ecosystem migrates to yet another build tool that nobody uses. It won’t help | 07:35:25 |
 danicheg | yeah, a sort of it https://github.com/scalacenter/scalac-profiling/pull/122 | 07:36:20 |
| matthiasberndt | If you use it, then it's not a build tool that nobody uses 😁 | 07:36:22 |
| anemish | You know what I mean. I don’t have anything against Mill. | 07:37:32 |
| matthiasberndt | Anyway, CanBuildFrom is gone, binary compatibility is much improved in Scala 3, so I think there are reasons to be optimistic. | 07:38:11 |
| anemish | Yeah, but the moment is gone as well. Now there’s Kotlin and Rust and TypeScript and even Java 21. Why Scala today? | 07:40:32 |
| matthiasberndt | I'm not going to argue that here | 07:43:24 |
| matthiasberndt | * I'm not going to argue that here, this is the Scala discord server. If you don't want Scala, go somewhere else | 07:43:55 |
| anemish | Oh I want Scala, I use for many years. It just hurts to see all this. | 07:45:08 |
 .zzyzzyxx | Moving away from the sbt topic here, but the short version for me is that Scala doesn't really compete with non-JVM languages like Rust and TypeScript, and among the JVM languages it easily has the best type system | 07:46:14 |
| anemish | Anyway, my point is that I believe it’s more important to try to fix sbt rather then mitigate to whatever | 07:46:23 |
| anemish | Yet Scala has the worst interoperability between JVM languages unfortunately | 07:47:51 |
| matthiasberndt | It does compete, Scala.js is absolutely a viable option these days | 07:48:14 |
| matthiasberndt | And native is getting closer | 07:49:15 |
| .zzyzzyxx | I suppose I was excluding what I view as "extensions" like Scala JS and Scala Native. If those were core goals by the language designers I'd feel differently | 07:49:43 |
| matthiasberndt | In fact, I think Caprese is going to make native a whole lot better | 07:50:19 |
 velvetbaldmime | They sort of are - Scala.js support is inside Dotty itself, and there is work at EPFL that targets Scala Native exclusively. | 07:50:24 |
| .zzyzzyxx | Well that is (good) news to me | 07:51:08 |
| velvetbaldmime | Both targets will also become a lot more first-class citizens in sbt itself, starting from sbt 2.0. The road ahead is long, but it's no longer an experimental phase - we're in ecosystem bootstrap phase | 07:51:13 |
| velvetbaldmime | In fact I believe it was Odersky himself who proposed having SNIPs - Scala Native improvement proposals 🙂 https://github.com/scala-native/improvement-proposals/pull/1 | 07:51:45 |
| velvetbaldmime | Anyways, this is getting wildly off topic | 07:51:55 |
| velvetbaldmime | W.r.t. SBT vs Mill - there doesn't have to be 1 single tool for ecosystem to be healthy. In fact, there's friendly co-existence and ideas borrowing: https://github.com/sbt/sbt/discussions/7180#discussioncomment-5364624 Java ecosystem is not limping along, and it has both Maven and Gradle JS ecosystem.. let's not go there, but plenty of options Python ecosystem is notorious for the amount of options, it's just a lot of them suck at the fundamental problem - whereas both Mill and SBT solve the fundamental problem (dependencies and management thereof) very well, because they both rely on Coursier. sbt's existence is marred by its startup performance, and eventually that will be solved. The quickest way to solve this would be to pay one of Scala consultancies to find and put a dedicated full-time person on solving that. But that's a lot of money. | 07:56:09 |
| velvetbaldmime | If anyone here wants to make concrete suggestions - do so politely and with respect to maintainers at https://github.com/sbt/sbt/discussions/7174 And note that there is a yet unresolved question of where the line is drawn - what improvements can become part of SBT 1.x, and what needs SBT 2.0 | 07:57:24 |
| anemish | Yes, that's what I was thinking of – chip in as an industry and dedicate a person to fixing sbt performance issues. I'm paying for that | 08:03:02 |
| velvetbaldmime | There have been precedents of companies loaning their employees to Scala Center full time, but I don't believe this ever covered a particular project. If such project can be drawn up with Eugene's help and guidance (like google summer of code, but for very senior people who love heap dumps and flamegraphs), I believe it will be possible to find a company to pay for that. | 08:05:04 |
| velvetbaldmime | Quickest way of course is for a company on the advisory board to bring this up and push it through, I don't think there will be much resistance. Just need to have a company with interest and seat on the board. I no longer work with Scala fulltime so can't really do anything directly, but I'll bring it up | 08:06:14 |
| anemish | Yes, that would be awesome. So, VirtusLabs, Software Mill, who else could do that? | 08:08:00 |
| anemish | Just imagine the rippling effect on the whole community! | 08:08:49 |
| anemish | Every Scala developer, project, company would benefit from it! | 08:09:28 |