miniboxing-plugin
miniboxing-plugin copied to clipboard
Clarify project status on the website and in the repo
It seems like the plugin is not developed anymore - were there any plans to utilize it in 2.12 or Dotty?
Thanks for the ping @clayrat! Indeed, I have found less and less time to work on miniboxing. Do you need it ported to 2.12 for a project?
Re. dotty, it will have an improved version of specialization based on the dotty linker, and you can see a paper on it by @DarkDimius here: https://infoscience.epfl.ch/record/217276?ln=en
No, nothing critical, I was just curious about porting old examples, e.g. https://github.com/tixxit/optimistic-spec-article
I guess, let's wait for Dotty then :)
I've updated the issue description to clarify the project status, so anyone looking to use it can contact me if they need support for 2.12.
Is the project dead thus? What about Scala 2.13?
It just needs resuscitation from a few kind-hearted contributors ;)
fyi @jorokr21. I've only seen scattered remarks about why the Scala 3 team hasn't pursued miniboxing further.
One such remark is Vlad's remark above about Dmitry's plans.
I also found https://twitter.com/alexelcu/status/1365162988362203136, which isn't very specific, but does kinda sound like @alexelcu asked somebody important about it.
Miniboxing isn't mentioned at https://contributors.scala-lang.org/t/status-of-specialization-in-scala-3/4628/10 (“Status of specialization in Scala 3”), but that thread lacks any real full response from an insider. There's also https://contributors.scala-lang.org/t/on-demand-specialization/4350/2 where Martin says "The best bet is probably to compile and specialize from Tasty"
Thanks everyone for your interest in miniboxing and thank you, @SethTisue, for following up.
TL;DR: As much as I would love to see miniboxing to Scala 3, this will need someone rolling up their sleeves and implementing it (and I don't have the resources to do this atm). The solution @odersky suggested in https://contributors.scala-lang.org/t/on-demand-specialization/4350/2 is very attractive, and I don't see any obvious drawbacks.
If anyone wants to give this a stab, please read the papers at http://scala-miniboxing.org/mbox/index.html and http://scala-miniboxing.org/ldl/index.html and ask questions here -- I'm following the thread and will answer to the best of my knowledge.
see also https://github.com/lampepfl/dotty/issues/15532