Falcon5757 Snoovatar
Falcon575723:32
We are still discussing how should we call it
See, the generic idea of a low level lisp specification (non as hoc sbcl) sounds tempting to many people, especially if you then say "well it should of course be such that the current CL standard would be implementable on top of it with no performance loss"
But there then should be reasoning about why this is not what racket is etc
(non ad hoc sbcl) jezz I cannot correct messages in here
flaming_bird23:43
"well it should of course be such that the current CL standard would be implementable on top of it with no performance loss" is a tough statement to make
since accounting for all of CL performance catches is troublesome and requires a lot of foresight, and likely getting existing implementers on board because they have the required experience
Falcon5757 Snoovatar
Falcon575723:44
It is tough
But if it wasn't that tough there's a good chance it actually might've been just racket
with a few quirks
flaming_bird23:45
I need to run off and do a thing now - be back tomorrow
Falcon5757 Snoovatar
Falcon575723:45
yes, the idea requires like, a lot of people and resources(
flaming_bird23:45
thanks for letting me know, I'm curious where you end up with this
you could try setting up an IRC channel on freenode for live chat
Falcon5757 Snoovatar
Falcon575723:46
I m not that familiar with IRC
flaming_bird23:46
like, dedicated to this issue
Falcon5757 Snoovatar
Falcon575723:46
I m not that familiar with a lot of things inclduing compiler building and theoretical CS
that's why i started it in chat and not alone
i need at the very least a huge document regarding the issue with a lot of useful info
flaming_bird23:47
I could do the paperwork and register the channel with Freenode and give owner rights to whoever is proper - just ask your Russian people if that's what they'd like to have
I mean, I've done that a few times, so I'm kinda on good terms with freenode channel service
Falcon5757 Snoovatar
Falcon575723:47
otherwise it ll get dismissed by actual compiler developers and that would be the end of it
flaming_bird23:47
but a channel is useless without people on it
"dismissed" is kind of a wrong word
Falcon5757 Snoovatar
Falcon575723:48
true and thats why for now i m good with just my russian chat and a couple of dms
flaming_bird23:48
it's also not once-and-for-all
even if at first the proposal ends up getting dismissed, you/we/people can figure out what's wrong with it and just fix it
and then go for round n+1 of that
like most big things in life, it's iteration, rather than one-shot
I'd also remember that a specification is worth little without a reference implementation
hence I'd try to create something that implements the semantics in the correct way, even if it's slow and not performant
something that can execute such code and can be fixed as appropriate when it doesn't conform to the specification
and can then become the oracle when it comes to the language semantics.
I see two choices: either write that in portable CL and execute this on top of any conforming CL impl
or hack an existing CL implementation until it no longer conforms to ANSI CL but instead conforms to this future CL you imagine
I'd prefer the former because it would run almost everywhere and make use of the existing CL libraries, but the latter might be more feasible for the beginning