[ruby-core:113670] [Ruby master Misc#19693] Data initialization is significantly slower than Struct

Issue #19693 has been updated by Eregon (Benoit Daloze).





Defining a `new` singleton method on the Data subclass, in Ruby and with ex=
plicit keyword arguments should be able to then use the literal kwargs opti=
mization, but that means using `eval` or so and then needing to validate al=
l argument names are valid local variable names (might already be the case,=
 I don't know). Or a way to define a method in C with the same kwargs optim=
ization, but that is currently not possible AFAIK.



A more general optimization of keyword arguments could also work for this, =
like https://github.com/oracle/truffleruby/issues/2388 is an example of tha=
t, but it's not fully implemented yet.



Another way is splitting `new` at every call site (for Data subclasses at l=
east) + escape analysis + a Hash representation that can take e.g. 10 pairs=
 and still escape analyze fine, then it should be the same as positional ar=
guments if everything inlines (should be, it's not big methods). I'll try t=
his on TruffleRuby when Data is implemented there.



----------------------------------------

Misc #19693: Data initialization is significantly slower than Struct

https://bugs.ruby-lang.org/issues/19693#change-103315



* Author: janosch-x (Janosch M=FCller)

* Status: Closed

* Priority: Normal

----------------------------------------

Maybe there is potential to make it as fast as Struct?



```ruby

require 'benchmark/ips'



S =3D Struct.new(:a, :b, :c, :d, :e, :f, :g, :h, :i, :j)

D =3D Data.define(:a, :b, :c, :d, :e, :f, :g, :h, :i, :j)



Benchmark.ips do |x|

  x.report('Struct') { S.new(1, 2, 3, 4, 5, 6, 7, 8, 9, 10) }

  x.report('Data')   { D.new(1, 2, 3, 4, 5, 6, 7, 8, 9, 10) }

  x.compare!

end; 1



# =3D> [...]

# =3D>             Struct:  6916530.4 i/s

# =3D>               Data:  1507259.5 i/s - 4.59x  slower

```







--=20

https://bugs.ruby-lang.org/

 ______________________________________________
 ruby-core mailing list -- [e-mail]
 To unsubscribe send an email to [e-mail]
 ruby-core info -- https://ml.ruby-lang.org/mailman3/postorius/lists/ruby-c=
ore.ml.ruby-lang.org/