libera/#commonlisp - IRC Chatlog
Search
13:24:44
Shinmera
somewhat confusingly, NIL is of type NULL, even though NIL is also a type (designating the empty type set).
14:14:50
jackdaniel
if T is a supersupertype, then it is only natural that nil is on the other end of the graph
14:19:04
Josh_2
All the lisp benchmarks are gone from https://programming-language-benchmarks.vercel.app/ :sob:
14:27:29
Josh_2
I dont think they are working on latest SBCL, the fella who has been writing the submissions has a few issues on the submissions repo
14:33:37
Guest7447
I'd try and argue you need to get people into the store before they buy, but then there'd be more questions of 'how do I make this work like x'
14:37:01
semz
Guest7447: Microbenchmarks like these would be false advertising under this analogy because they get gamed immediately. The Computer Language Benchmark Game is a great example: The top scoring submissions are usually written in inline assembly rather than the language in question.
14:37:04
Guest7447
it's marketing, good or bad. it just seems whenever a person has 'heard lisp can be fast' they ask questions about how optimize things rather than how to write Lisp.
14:37:52
semz
The most egregious example is probably the Python regex benchmark (which gets place #2 or so by using the foreign function interface to use a completely different regex library instead)
14:38:57
Guest7447
though things like that still show a) what's possible b) how it's done. I think it could be good marketing to show off simd use in lisp.
14:47:36
Josh_2
An excessive concern with "performance" is normally an indicator that this person has been consuming large quantities of soy
15:00:45
shka
one of problems with benchmark game was that java and C++ were putting out multi-thread programs
15:02:05
shka
and the only thing that you can get out of those is that there is a distinct performance tier for static typed compiled languages with a light to moderate optimizations employed
15:04:28
shka
anyway, rule of thumb, if you think that something can be done efficient enough in java, it can be done with SBCL as well
15:05:29
rotateq
trev: Ah okay, so you're experienced yet I guess. But I hope you don't think SICP is good for pure beginners.
15:05:38
trev
and looked for a LISP to use...started reading and comparing, then decided that CL is The One
15:06:42
rotateq
If you want to really use the code from newer version of SICP, this is good with Racket. But yeah, CL is the way.
15:06:51
trev
rotateq yeah, SICP is definitely not, but i am providing assistance. I think it does a decent job with foundational knowledge, moreso than the Python book she is reading
15:07:36
trev
rotateq, yeah exactly! i had her use Racket for the examples from the book. CL is my own interest
15:13:35
trev
Racket seemed interesting with the "language oriented programming", but i am not really sold on that paradigm
15:14:21
rotateq
Well we have that too, you have a programmable programming language and can fit it to the needs of a problem.
20:58:51
White_Flame
when outside of macroexpansion and in normal runtime code, what's the preferred way to refer to the package that the source code is written in? Does (intern str #.*package*) make sense for all cases?