@nomeata@mastodon.online
@nomeata@mastodon.online avatar

nomeata

@nomeata@mastodon.online

Has a thing for abstraction.
Haskeller, Computer Scientist.
Dances tango, swing and blues.
Stand-up comedian and Paraglider.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

frommMoritz, to random German

The PR Tracker by @qyliss is just such a neat tool!

https://nixpk.gs/pr-tracker.html?pr=226169

nomeata,
@nomeata@mastodon.online avatar

@frommMoritz @qyliss I wish such auxiliary tools were easier to discover. A bot posting a link to every PR is surely too noisy. And too bad one can't configure GitHub with a bunch of extra per-PR links in the side bar.
Maybe a “Github status” that links to that page? Less visible once merged, however.

nomeata, to programming
@nomeata@mastodon.online avatar

I’m (experimentally) extending the hs-all-in-one tool to munge the whole GHC source code into a single module. I now have a 680k loc, 41MB file. Doesn’t work yet, still fighting the renamer.

Initial motivation is GHC, picking up thoughts from <https://www.joachim-breitner.de/blog/748-Thoughts_on_bootstrapping_GHC>.
The most viable boostrappable Haskell compiler is Hugs (written in C) but doesn’t support recursive modules. So maybe if all of GHC is one module it could handle it…

nomeata,
@nomeata@mastodon.online avatar

@terrorjack Right now I am irrationally trying to get the “All of GHC in one file” thing going, just for the sake of it. Even if only GHC can compile it at first.

This is of coufse far from being the only problem towards bootstrappability.

nomeata,
@nomeata@mastodon.online avatar

@terrorjack
My hope (which I need to investigate) is that hugs can maybe compile haskell-src-exts and haskell-names. This means we can write simple preprocessors (such as combining recursive modules) in Haskell.

Then we’ll see which extensions ghc actually uses, which we can simulate, which can teach to Hugs, and how far we’d have to go back in history.

nomeata,
@nomeata@mastodon.online avatar

@terrorjack According to a very brief glance, the first use of GADTs in GHC are nitroduced by Simon Marlow in 2011, in this commit https://gitlab.haskell.org/ghc/ghc/-/commit/889c084e943779e76d19f2ef5e970ff655f511eb so we might go back to 7.0.4

(Or maybe later and then disable the code generator in the build flags.)

nomeata,
@nomeata@mastodon.online avatar

@terrorjack The question is when things are getting so old that it becomes harder again – incompatibilities with current C toolchain for example, as I ran into with my brief experiments in https://gitlab.haskell.org/ghc/ghc/-/commit/889c084e943779e76d19f2ef5e970ff655f511eb

nomeata, to random
@nomeata@mastodon.online avatar

I'm wondering if I should package and contribute little useful (at least to me) but somewhat fringe tools like git post-squash to nixpkgs:
https://github.com/nomeata/git-post-squash

xavier, to NixOS

I'm looking into , which I know ~nothing about, and I just discovered that FaaS "functions" are essentially Docker images, which seems pretty heavy to me.

This makes me wonder whether it would make sense to build an FaaS framework on top of , where "functions" are derivations that produce an executable, which is run in a sandbox. I wonder if anyone's made something like that already? I couldn't find anything on Google.

@bryanhonof maybe you've heard of such a thing?

nomeata,
@nomeata@mastodon.online avatar

@xavier @bryanhonof on AWS Lambda I upload plain static executables, no Docker around. And Fastly has it even more fancy, supporting WebAssembly modules.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • megavids
  • thenastyranch
  • rosin
  • GTA5RPClips
  • osvaldo12
  • love
  • Youngstown
  • slotface
  • khanakhh
  • everett
  • kavyap
  • mdbf
  • DreamBathrooms
  • ngwrru68w68
  • provamag3
  • magazineikmin
  • InstantRegret
  • normalnudes
  • tacticalgear
  • cubers
  • ethstaker
  • modclub
  • cisconetworking
  • Durango
  • anitta
  • Leos
  • tester
  • JUstTest
  • All magazines
    •