Stas is a machine. Almost every single day he's in the sources making improvements. I joined the sbcl-commit list recently and the project is far more active and cared for than I had originally suspected.
stackghost 21 hours ago [-]
SBCL is a treasure. I really enjoy writing lisp, I just wish the library ecosystem was a little more mature.
I still use it whenever I can
CyberDildonics 19 hours ago [-]
It has only been 68 years, give it a little more time.
matheusmoreira 1 days ago [-]
> sequences are a band-aid over the lack of real iterator protocol
Wouldn't it be better to solve that problem? Proper generalized iteration with generators is just semicoroutines which can be implemented by swapping around stack pointers with zero copying. Should be competitive performance wise.
BoingBoomTschak 1 days ago [-]
As linked, the extensible sequence protocol (https://shinmera.com/docs/trivial-extensible-sequences/) already exists and is quite workable. The problem is that it's not supported enough, both by implementations (CCL missing, for example) and the entire ecosystem.
In the end, it's simpler to stay with boring, ANSI compliant solutions well optimized by compilers.
shawn_w 22 hours ago [-]
I've been toying with trying to add extensible sequences to ccl as a way to get more familiar with it, but ccl development seems dead so I'm not sure it's worth the effort.
stackghost 21 hours ago [-]
There are effectively two Common Lisp worlds: the commercial world where Allegro and Lispworks dominate, and the non-commercial world where SBCL is more or less the only game in town.
CCL, as far as I can tell, is abandonware
guenthert 12 hours ago [-]
> CCL, as far as I can tell, is abandonware
Last release was August 14 2024. For a 30+ year old project, that's quite recent methinks.
It clearly doesn't move as fast as SBCL, but I wouldn't call it abandoned either.
ototh, afaiu, SBCL gave up on ARM32. Can't blame them for that, but until 2023 I was still using an early Banana Pi with such. CCL worked there much better (also it's GC seems more robust).
a-french-anon 16 hours ago [-]
Personally, I still try to support CCL, ECL and CLISP in addition to SBCL; I mostly ignore CLISP's warnings though.
tmtvl 12 hours ago [-]
ECL still has its niches (if you want to use CL in an Android application, for example, ECL is easier to work with than SBCL).
14 hours ago [-]
tug2024 19 hours ago [-]
[dead]
ezoi_MuUJd 23 hours ago [-]
[flagged]
Rendered at 23:54:04 GMT+0000 (Coordinated Universal Time) with Vercel.
I still use it whenever I can
Wouldn't it be better to solve that problem? Proper generalized iteration with generators is just semicoroutines which can be implemented by swapping around stack pointers with zero copying. Should be competitive performance wise.
In the end, it's simpler to stay with boring, ANSI compliant solutions well optimized by compilers.
CCL, as far as I can tell, is abandonware
Last release was August 14 2024. For a 30+ year old project, that's quite recent methinks.
It clearly doesn't move as fast as SBCL, but I wouldn't call it abandoned either.
ototh, afaiu, SBCL gave up on ARM32. Can't blame them for that, but until 2023 I was still using an early Banana Pi with such. CCL worked there much better (also it's GC seems more robust).