Yeah, I had a series of points I wanted to make, but I was obliged to break it up, which unintentionally caused misunderstandings. It's difficult to conduct a long-form discourse on social media. Heck, it's difficult even to converse on social media, at all.😀
We all know that loads of people implement Scheme, especially the much simpler R4RS, using many different languages, including ML. But if I understood you right, you are considering using ML as the embedded macro processor for R7RS Scheme, yes?
Well, that's just brilliant! Indeed, ML was called "Meta Language" because it was born as such: an embedded, scripting language for the LCS theorem prover. So, you're using ML as it was originally intended, by employing it as a macro language.
The most popular ML implementations are SML/NJ (the gold standard), MLton (whole-programme optimiser), HaMLet, and Moscow ML. Of those, Moscow ML is perhaps the smallest, hence arguably the simplest. It compiles fast, but the generated code isn't fast. But that shouldn't detract from this particular use case. You will also be able to leverage MLlex and MLyacc.
Correct me, if my reading of your post was off.
DougMerritt (log😅 = 💧log😄) (npub1vlc…mzqq)
amen zwa, esq.
npub1gn…30jmx
2026-01-08 12:21:08 UTC
in reply to nevent1q…flvj
Author Public Key
npub1gnu7eyvs2nd4u32253956ywmvycdck9552wqe9cea2gf6k56uyqql30jmxSeen on
wss://relay.momostr.pinkPublished at
2026-01-08 12:21:08 UTCEvent JSON
{
"id": "0750f1d4d81dfa4d8dffcc8ada094a75633236087d5757a689815dd866c07a38",
"pubkey": "44f9ec919054db5e454aa44b4d11db6130dc58b4a29c0c9719ea909d5a9ae100",
"created_at": 1767874868,
"kind": 1,
"tags": [
[
"p",
"44f9ec919054db5e454aa44b4d11db6130dc58b4a29c0c9719ea909d5a9ae100"
],
[
"e",
"216e323cdb7707bdc928d6ddcf7312815cc7c3bc2df433700f120c49fd76f977",
"",
"reply",
"db6af3fbdb028e1ec5857bb309c8686dde3b884fe8243bfa8b302a0875ab53d7"
],
[
"e",
"6cbb10b2160895959b7d9740e4b9ed9ed3298893b1c1a5cd7da2873a5e5c4c87",
"",
"root",
"44f9ec919054db5e454aa44b4d11db6130dc58b4a29c0c9719ea909d5a9ae100"
],
[
"proxy",
"https://mathstodon.xyz/@AmenZwa/115859447360143612",
"web"
],
[
"p",
"db6af3fbdb028e1ec5857bb309c8686dde3b884fe8243bfa8b302a0875ab53d7"
],
[
"p",
"67f17e59b525a7cc73f7a2bb4308053daa9c4db30d42b9fcb237903d4a93043f"
],
[
"proxy",
"https://mathstodon.xyz/users/AmenZwa/statuses/115859447360143612",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://mathstodon.xyz/users/AmenZwa/statuses/115859447360143612",
"pink.momostr"
],
[
"-"
]
],
"content": "Yeah, I had a series of points I wanted to make, but I was obliged to break it up, which unintentionally caused misunderstandings. It's difficult to conduct a long-form discourse on social media. Heck, it's difficult even to converse on social media, at all.😀 \n\nWe all know that loads of people implement Scheme, especially the much simpler R4RS, using many different languages, including ML. But if I understood you right, you are considering using ML as the embedded macro processor for R7RS Scheme, yes?\n\nWell, that's just brilliant! Indeed, ML was called \"Meta Language\" because it was born as such: an embedded, scripting language for the LCS theorem prover. So, you're using ML as it was originally intended, by employing it as a macro language.\n\nThe most popular ML implementations are SML/NJ (the gold standard), MLton (whole-programme optimiser), HaMLet, and Moscow ML. Of those, Moscow ML is perhaps the smallest, hence arguably the simplest. It compiles fast, but the generated code isn't fast. But that shouldn't detract from this particular use case. You will also be able to leverage MLlex and MLyacc.\n\nCorrect me, if my reading of your post was off.\n\nnostr:npub1vlchukd4yknuculh52a5xzq98k4fcndnp4ptnl9jx7gr6j5nqsls06mzqq",
"sig": "9522efda6460a11807b40f88f65982b9709fbfb9a31aa754008714fe785e063b10e51619362b83606b7a524b2167db97039c900705d79d3c9a33006cd790499d"
}