Hi @spdegabrielle I am confused about your question.
How is Google supposed to provide a service to a user without keeping some information about that user? At the very least, they would need to store the email address used to send messages from the racket-users list. And this is what an account really is. How is Google supposed to provide the service without the user having an account on their system?
Or is this question about (a) Having to use the web interface to sign up? (b) Having to use a gmail address to sign up?
A user reported being unable to sign up to racket-users on reddit. They specifically mentioned not wanting to use a google account
Samph was able to confirm that you can sign up to the mailing list without using or creating a g-acc by emailing the subscribe, and replying to the response
It’s my plan to create pr for https://lists.racket-lang.org/\|https://lists.racket-lang.org/ to make this clearer & easier for new subscribers.
@alexharsanyi I think there’s a cultural expectation in the world of mailing lists that the only thing you should need to participate in them is an email address, and it shouldn’t matter what that address is or who’s hosting it. I think that’s a good thing, personally.
I just need to review announce and dev, as well broken links including the archive links (gmane is dead).
And you CAN use it without a goog-account - but how to do this isn’t clear on the lists page.
@narendraj9 has joined the channel
why is the toplevel hopeless?
@spdegabrielle Is that page linked to in the Wiki?
I was hoping someone (not me:sob:) could write a paragraph and use one of the examples in the gist to make a wiki page
Actually this is probably better @hazel :
Thank you!
The short, very informal answer is that, “the top level is hopeless” because the REPL doesn’t have all the information it needs to correctly compile code, since it only gets the code in chunks.
A module compiles correctly because all of the code can be analyzed at the same time, so the compiler can figure out how exactly what the different parts are and how they’re supposed to interact.
I feel like the repl needs a short preamble that explains it. (Just after the Lang & version stuff)
(Hopefully my tl;dr wasn’t too far off base.)
I’m assuming strategies like disallowing binding forms in the REPL so users can’t get into this situation is too aggressive ?
yeah, the repl is very useful, it’s just impossible to give semantics to
Several languages support incremental evaluation. Some even have macros similar to Racket (Lean 4). So I think the problem should be fixable in Rhombus.
My guess is that what we would lose is implicit mutual recursion and stuff like syntax-local-lift-module-end-declaration? Explicit mutual recursion is a better idea to me anyway, though HtDP might not like this change.
It’s certainly possible to create REPLs with sensible semantics, but not while preserving the traditional LISP behavior. ML has had a REPL for a long time, for example.
What particular feature of LISP REPLs makes it so hard to pin down the semantics?
I do recommend reading the linked messages, but it’s the combination of forward reference with macro expansion
I think there is a missing (register-resolved-type-alias name (make-Name name (length tvars) #t) call in tc-toplevel-form that might make this work. Maybe there is some reason it wasn’t added in the big toplevel refactoring, but might just be a mistake.