News for hackers. Stuff that matters.
Re:Racket v7.5 by Buttons840 on Wednesday November 20, 2019 @11:46AM (#21585517)
> Racket CS remains "beta" quality for the v7.5 release, but its compatibility and performance continue to improve.

Does this mean Racket CS is faster than the default Racket?

Re:Racket v7.5 by nikofeyn on Wednesday November 20, 2019 @12:16PM (#21585894)
sometimes yes, sometimes no.

https://blog.racket-lang.org/2019/01/racket-on-chez-status.h...

it seems it's a bit 50-50 in terms of performance gains, but the main gain, from what i understand, is the ability for the racket developers to code in racket and chez scheme now instead of c.

Re:Racket v7.5 by gus_massa on Wednesday November 20, 2019 @02:24PM (#21587451)
Note that this report is 9 months old. I guess there will be an new report this January. There has been a lot of small improvements.

I'm not tracking all of them, but I think that now there are less checks to ensure that the functions return don't return multiple values in the wrong moment, and there are less unnecessary continuation marks.

Re:Racket v7.5 by velcrovan on Wednesday November 20, 2019 @12:19PM (#21585946)
It does not mean that. It just means Racket CS is faster in 7.5 than it was in 7.4.
Re:Racket v7.5 by igouy on Wednesday November 20, 2019 @10:56PM (#21591310)
Re:Racket v7.5 by peatmoss on Thursday November 21, 2019 @10:13AM (#21594824)
As Racket CS slowly improves towards matching the mainline implementation, those results can give the impression that Racket CS is slow. To put things in context, look where Racket was already at in comparison to another very dynamic language, Python:

https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

I don't know how to directly compare Python 3 and Racket CS on that site, but comparing three ways paints Racket CS in a more favorable light. Hopefully Racket CS will close the gap with Racket, but it’s already doing pretty great relatively... unless we compare to SBCL, Rust, Haskell, etc.

Re:Racket v7.5 by soegaard on Friday November 22, 2019 @05:33AM (#21604426)
Note that several of these benchmarks use 4 cores on Python but only one on Racket. Time to revisit the Racket programs and use places.

Look at the single core benchmarks to get a more 1-1 comparison.

Re:Racket v7.5 by igouy on Friday November 22, 2019 @01:34PM (#21608474)
> Time to revisit the Racket programs and use places.

Please contribute improved programs:

https://salsa.debian.org/benchmarksgame-team/benchmarksgame/...

Re:Racket v7.5 by igouy on Friday November 22, 2019 @10:02AM (#21606127)
Please note the URL I provided shows a comparison between Racket and Racket CS (not Python).