Nathan Faubion

[Nathan Faubion @natefaubion]

Just published v1.3.0 of purs-backend-es, which has some goodies for writing lower-level imperative code.
https://t.co/o3NBs69LTs

[Nathan Faubion @natefaubion]

@paf31 But can it run Loom?

[Nathan Faubion @natefaubion]

@drathier @evanrelf I'm not sure how the erl backend compiles pattern matching, but in this IR there is no native case, it is compiled to boolean predicates and optimized.

[Nathan Faubion @natefaubion]

I am extremely excited to announce a new #PureScript project from Arista NDR: `purescript-backend-optimizer`
https://t.co/zHZyA1pgNn A new optimization pipeline and alternative modern-ES backend for PureScript!

[Nathan Faubion @natefaubion]

@drathier @evanrelf It is not a CoreFn to CoreFn. CoreFn is too high level to support many of the optimizations you'd want. It's a slightly more mid-level IR with built-in things for effect binds, uncurrying, primitive operators, etc.

[Nathan Faubion @natefaubion]

@drathier @evanrelf The optimization pipeline is totally backend agnostic. You could write a purerl generator that uses it. I think purerl is written in Haskell though, so you’d either need to write a new erlang code generator in PS or serialize the AST.

[Nathan Faubion @natefaubion]

@evanrelf If you dig deep enough in the repo history, this started as an example Lua backend 😂

[Nathan Faubion @natefaubion]

@goodacre_liam I assume you have some dark magik prepared for such an occasion?

[Nathan Faubion @natefaubion]

@GabriellaG439 The cognitive dissonance of your state dish having so many Qs and Ts must be real.

[Nathan Faubion @natefaubion]

@GabriellaG439 You spell brisket with a Q? Jail.

[Nathan Faubion @natefaubion]

@GabriellaG439 Presumably if you still allow a monadic interface with some RTS you will still have the issue of type errors interleaving with effects.

[Nathan Faubion @natefaubion]

In light of recent Haskell discussions.

[Nathan Faubion @natefaubion]

@hdgarrood Implicit parens when substituting introduce rank-2 forall. Make parens required in type synonyms with foralls so this is explicit.

[Nathan Faubion @natefaubion]

@hdgarrood Implicit parens when substituting introduce rank-2 forall. Make parens required in type synonyms with foralls so this is explicit.

[Nathan Faubion @natefaubion]

Pure FP kinda sucks for PL implementations, specifically Haskell. It’s all bags of mutable state and graphs, which are unnecessarily inefficient when modeled purely.

[Nathan Faubion @natefaubion]

@GabriellaG439 I feel like this blog post is extremely generous to it's own argument in assuming that the Python community would have voiced less annoyance given more frequent cross-cutting breaking changes. At the scale of Python, all breaking changes incur wrath.

[Nathan Faubion @natefaubion]

@hdgarrood @GabriellaG439 Now multiply this by around 100 packages (immediate packages that everyone relies on). Moderately batched breaking changes are by far preferable. I feel like the blogs take fails to address the optics of "I can't keep up with the churn anymore" reaction.

[Nathan Faubion @natefaubion]

@taylorfausak I feel like this is asking someone to publish a really long package name that’s only consonants.

[Nathan Faubion @natefaubion]

@artemisystem @GabriellaG439 I think it's named that way because it's like the "lowered" version of Alternative's guard. `a <|> guard p $> b` vs `a <> guard p b`. They are both monoidal.