[23409125]

Subpar performance is a huge drawback of Perl 6.

I usually don't program in Perl, but occasionally use it instead of sed.

Compare, Perl 5 vs. Perl 6:

    $ time yes | head -n1000000 | perl -pe 's/y/n/' >/dev/null

    real	0m0.945s
    user	0m0.944s
    sys	0m0.016s
    $ time yes | head -n1000000 | perl6 -pe 's/y/n/' >/dev/null

    real	2m49.881s
    user	2m44.892s
    sys	0m2.184s

Spending several minutes to do what can be done in under 1 second is just unacceptable.

[artagnon]

It's nice to dream up a language with the perfect semantics; in reality, all programming languages make compromises because they are limited by current compiler technology. Ruby is on one extreme end of the spectrum, making little compromises on beauty, but we all know how slow it is.

Go is on the other end of the spectrum; they have tuned their syntax to be fast to parse, and have stuck to implementing existing compiler technology. If your semantics are very far removed from how computers work, you need a very advanced compiler like GHC.

It sounds like Perl6 is somewhat centrist in its approach, and has innovated quite a bit; I think it will be possible to almost match Perl5's speed with reasonable effort. The interpreter might do well on small examples, but the VM will dominate in very large ones. Many compilers are tiered in that they use an interpreter for cold code, and spend the effort to generate code and JIT hot code.

[zoffix222]

Yes, performance is where most of the funding is currently going into.

Bleed commits on VM for example, recently passed Perl 5 on speed with read-a-million-line-file-and-sum-up-number-of-chars bench ( https://twitter.com/zoffix/status/895684203550973956 ).

[23409125]

What are bleed commits?

[jacobolus]

The “cutting edge” of technology or craft is the newest / greatest stuff, pushing the field forward in new unexplored directions, by analogy to the cutting edge of a plow or a knife which slices through some uncut material or unbroken ground. I believe the metaphorical use of the term dates from the middle of the 20th century.

The “bleeding edge” is a more recent term for the same idea, emphasizing that being at the cutting edge might involve danger to yourself.

[zoffix222]

Commits that happened recently and aren't part of any release yet :)

[colejohnson66]

Commits that would go into the bleeding edge builds

[wott]

Yes, I haven't tried lately, but for my own small text manipulation program, Perl 6 was still more than 100 times slower than Perl 5 the last time I tried. Probably because of some JIT delaying the startup time. But since my use is calling small programs, it was a no go for me; and after one year it had not improved a bit.

[visarga]

Many string and dictionary based operations seem to work faster in Perl5 than Python for me.

[Steeeve]

Perl5 string and hash performance was always top notch.

[wott]

The Perl 5 version was indeed also 3 to 4 times faster than the Python one.

[issaria]

Did you compare to python2 or python3, cause python3 is in the same league as perl6.

[logicallee]

wow, that is insane. You picked an awesome example.

We're only talking about on the order of a million bytes in your example, so 2 minutes to process on the order of 1 megabyte with a regex is just insane. If you applied it to 1 gigabyte of files that would be an even 2 days vs Perl 5 @ 16 minutes.[1] Sixteen minutes isn't enough to rewrite it in C, so Perl 5 is "fast enough". You can go get a coffee or think about the next thing you want to do with the data, or write the next part of your processing. But two days is prohibitive: it breaks your workflow.

[1]

https://www.google.com/search?q=2+minutes+49+seconds*+1024

https://www.google.com/search?q=0.94+seconds+*+1024

[throwaway7645]

It might help to point out that the core team only froze the spec during Christmas last year or year before and have therefore just started on performance in a way. If you check the perl6 weekly site you'll see significant progress on a weekly basis (ex: Lists sped up by 1500%).

[ash_gti]

So, a fairly simple change to this will bring down the time to a few ms (86 ms on my computer)

  time yes | head -n1000000 | perl6 -e 'for $*IN.readchars { .subst("y", "n").print }’

Its slightly more complicated because I’m having it read the input in a chunk instead of on a per-line basis and I’m using subst on a literal instead of a regex.

[davemp]

> \d now matches Unicode digits such as ᭕ (a Balinese five) and ๓ (a Thai three).

This sounds like a disaster waiting to happen.

[perlancar3]

It has been like that since forever (at least since 5.10, 2007). To only match 0-9, use [0-9] or /\d/a (/a is available since 5.14, 2011).

[ab5tract]

How, exactly?

[cestith]

Perl 6 is a spec. The current Rakudo implementation is being focused toward sped now that the spec is stable.

[gigatexal]

did you file a bug? is there a way to trace the perl6 one and see where the hotspots are?

[perlgeek]

perl6 --profile yourprogram.p6

[smegel]

It is a drawback of the current implementations of Perl 6, yes.