CMUCL is actively developed

[fehmud]

Why is CMUCL listed "mainly for historical purposes"? It looks like it is actively developed: the last release was a little more than one year ago, and the latest commit in its source repository is from one week ago. Thank you.

[Hexstream]

Probably because SBCL is assumed to be vastly better than CMUCL for almost all use cases.

[fehmud]

Probably because SBCL is assumed to be vastly better than CMUCL for almost all use cases.

But for that there are already the ⭐ and 👍 icons, aren't there?

[Hexstream]

Yes, but there is no 👎 icon. ;)

[vindarel]

I am surprised to see all this activity (https://gitlab.common-lisp.net/cmucl/cmucl/-/commits/master) (some commits pushed in April dating from 2018) (https://gitlab.common-lisp.net/cmucl/cmucl/-/issues)

We can't by any means encourage folks to try CMUCL. But we don't say it is "dead" so all is well ;)

Take CLISP and Corman Lisp also listed for "historical reasons": they see some activity but not much, they are lagging behind, they are not 100% ANSI compliant. But they both have a lil' interest so we leave them here. Note that we didn't list all historical CL implementations.

[Hexstream]

The unofficial fork of CLISP is seeing some activity, CLISP itself has not seen any for nearly 15 years.

(I really wish that fork would just rename itself to "KLISP" or something to put an end to all this stupid confusion.)

[daninus14]

they are not 100% ANSI compliant

That's not an argument. There's no implementation which is 100% ANSI compliant (until we get SICL) since they all use MIT Loop which is missing features described in the spec...

[Hexstream]

There's no implementation which is 100% ANSI compliant

Which specific features do you claim SBCL is missing, for instance?
They claim to be fully compliant and I certainly trust them on that.

[daninus14]

http://metamodular.com/SICL/loop.pdf

The following works in SBCL 2.4.9:

(loop until (> i 20)
  for i from 0
  do (print i))

And according to the spec it shouldn't.

[Hexstream]

Ok, well certainly feel free to report this to the SBCL team, but I just don't think a few super obscure bugs like that somehow means that blatantly non-conforming implementations (official CLISP comes to mind) are on exactly the same level of conformance as nearly 100% conformant implementations such as SBCL...

(Also, LOOP should die a horrible death anyway, and I'll personally make sure it does in a few years. ;P)