@abcdw Looks fantastic!

...but where is scrumble egg?

@abcdw gees, man that's a lot of food right there 😁

I do my 6-minutes typing workouts in between other tasks, so my brain get confused which of the activities is work and which is rest, so I mostly recieve a feeling of pleasure and satisfaction out of typing practice.

@abcdw Markiplier's tasteful nudes calendar but it's Linus Torvalds

@abcdw my legacy lives! :ablobcatattentionreverse:
Also rail sounds very interesting, exactly what I wanted

@abcdw I am gonna try it today.

@abcdw for ports in general I use (rnrs io ports), but with exceptions you get the weird the r6rs error messages, which while fine on their own & r6rs providing the superior conditional system, they are disjoint from Guile's style which feels a little clunky

I went with (scheme base) aka r7rs for binary port and (ice-9 exceptions) aka guile exceptions, but this fragmentation of ecosystem is not cool at all and raises learning curve to the sky.

https://git.sr.ht/~abcdw/guile-nrepl/commit/fc2559710ce489d1e5922c6c5916675d67ee15bd

#scheme #guile #lisp

@abcdw My personal preference: (ice-9 binary-ports) if you’re writing Guile-only code, or (rnrs io ports) if you’re aiming for R6 portability. These interfaces are roughly synonyms; the R6 one just pulls in more stuff.

(I’m skeptical of R7, I think it didn’t start on the right foot.)

@abcdw

In TXR Lisp I invented an alternative way to dealing with the local, scope-tied resources of blocks of code that are being entered and exited by continuations. (Like say you have a recursive walk with local resources, which is using continuations to yield incremental results elsewhere.)

Bascially, I allow a Lisp form to be terminated by "absconding": a dynamic non-local return, without any unwinding. This allows continuation switches to be clean, like thread context switches.

@abcdw are you saying funroll loops doesn’t secretly add rainbows and glitter

@abcdw lies and slander

@abcdw Like being on Slashdot, back in the day.

@abcdw frontpage of hackernews can either be a blessing or a curse. So far the comments on this post seem complimentary of Guix, at least.

@abcdw the only thing I can think of right now is just being explicit about it. Returning a vector doesn't help anyone understand that the vector is of a fixed length, or technically a tuple.

oh actually I guess one thing I like about the ability to do it in Common Lisp is that unless you explicitly need the other return values you just get the first one. Although I guess that could be seen as a downside too.

anyway, that's just my speculation and opinion, I don't usually need or want it.

@abcdw I think this generally might be related to low-level optimization, like compiler being able to spill mutliple values into registers, as opposed to vector, which should be honest-to-god allocated in heap (or on stack).

@abcdw It is a game changer: "the Hoot-compiled WebAssembly ... is five times faster than native Guile!"

I feel like some amazing stuff has been done recently in the Guile world.

@abcdw

"The web should be for everyone. We hope Hoot paves a path so that more languages may enter the web, including Python, Lua, Haskell, ML, and friends."

That's what I'm eagerly waiting for :)