Re:NewLisp
by thegeomaster on Friday September 25, 2015 @05:28PM
(#10280809)
I was referring more to their startup time. The runtime introduces delay when you type in the script command on a CLI.
Regardless of that, the factor-of-two-or-sometimes-faster argument sounds a little far-fetched for me, care to provide some sources on the benchmarks? Not saying it's not true, I'm just not accustomed to seeing highly extensible and dynamically-typed languages such as Common Lisp perform so fast.
Check out the benchmark game for some decent benchmarks of various languages. Here's sbcl vs gcc c:
https://benchmarksgame.alioth.debian.org/u64q/benchmark.php?...
On several benchmarks sbcl is less than three times slower than c.
And I never noticed a startup delay with sbcl. Let's do a simple test.
$ echo 'print("hello")' > test.py
$ echo '(format t "hello")' > test.lisp
$ time python test.py
hello
real 0m0.033s # 33ms
user 0m0.027s
sys 0m0.003s
$ time sbcl --script test.lisp
hello
real 0m0.006s # 6ms
user 0m0.000s
sys 0m0.003s
# I couldn't figure out how to run newlisp w/o starting
# the interactive interp so I'll give it a string
$ time newlisp -e '(print "hello")'
hello"hello"
real 0m0.002s # 2ms
user 0m0.000s
sys 0m0.000s
So sbcl starts up about 5 times faster than python, which lots of people use for scripting. newlisp seems to startup about 4ms faster than sbcl. I really don't see a scenario where the startup time of sbcl is a problem. Just for fun let's look at java. $ java -version
openjdk version "1.8.0_60"
OpenJDK Runtime Environment (build 1.8.0_60-b24)
OpenJDK 64-Bit Server VM (build 25.60-b23, mixed mode)
$ time javac hello.java
real 0m0.386s # ouch
user 0m0.910s
sys 0m0.040s
$ time java hello
hello
real 0m0.066s # 66ms, once compiled
user 0m0.067s
sys 0m0.007s
Re:NewLisp
by PackageUpgrades on Friday September 25, 2015 @09:00PM
(#10281663)
sbcl is fast, and can be really fast. Especially compared to the other cl implementations.
Re:NewLisp
by PackageUpgrades on Friday September 25, 2015 @08:40PM
(#10281595)
Commercial Lisps are better? As in faster?
Do you have any supporting evidence of this?
Having used Lispworks and Allegro I can't see where this is coming from.
Say what now? SBCL is normally within a factor of 2 of C, and sometimes faster than C, and the commercial Lisps are even better.