Added jemallocator only for Linux targets

[Razican]

As per the list of supported targets for jemallocator, it seems that only x86_64-unknown-linux-gnu is properly supported with all tests passing. So I added a conditional dependency that will only build in those targets and will only use jemallocator then. This fixes #309.

[github-actions]

Benchmark for aec9f79

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	407.7±21.28µs	398.6±16.81µs	102%
Expression (Parser)	2.9±0.15µs	3.0±0.18µs	97%
Hello World (Execution)	437.1±22.61µs	444.0±33.57µs	98%
Hello World (Lexer)	1026.0±64.13ns	933.4±52.80ns	110.00000000000001%
Hello World (Parser)	1138.4±55.96ns	1065.1±58.09ns	107%
Symbol Creation	512.3±30.05µs	528.8±26.76µs	97%
fibonacci (Execution)	4.8±0.17ms	4.7±0.21ms	101%
	undefined	undefined	100%

[github-actions]

Benchmark for 1a9db43

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	458.5±24.46µs	453.1±24.25µs	101%
Expression (Parser)	3.4±0.21µs	3.4±0.22µs	100%
Hello World (Execution)	484.9±15.73µs	487.4±60.85µs	100%
Hello World (Lexer)	1034.7±49.06ns	1140.2±38.56ns	89.99999999999999%
Hello World (Parser)	1218.3±85.42ns	1243.6±72.74ns	98%
Symbol Creation	587.5±40.43µs	558.2±31.79µs	105%
fibonacci (Execution)	5.3±0.18ms	5.2±0.21ms	100%
	undefined	undefined	100%

[Razican]

Interestingly enough, benchmarks don't seem to be improving that much. Maybe it's not using jemallocator? I will check it in my computer.

[jasonwilliams]

This now builds fine on Windows

[jasonwilliams]

Interestingly enough, benchmarks don't seem to be improving that much. Maybe it's not using jemallocator? I will check it in my computer.

I'm not sure, maybe the triple isn't quite right?

[Razican]

Should be fixed now. It turns out there is no target in cfg, as per the reference, and we need to specify each part separately. Let's wait for the benchmarks.

[github-actions]

Benchmark for fddf015

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	325.5±23.17µs	380.2±24.33µs	83%
Expression (Parser)	2.6±0.24µs	2.9±0.22µs	86.00000000000001%
Hello World (Execution)	346.9±28.41µs	405.2±24.60µs	83%
Hello World (Lexer)	1046.9±154.52ns	863.7±51.76ns	121%
Hello World (Parser)	1098.0±87.26ns	967.7±53.91ns	112.99999999999999%
Symbol Creation	354.6±24.05µs	479.1±32.31µs	64.99999999999999%
fibonacci (Execution)	3.8±0.15ms	4.6±0.29ms	78%
	undefined	undefined	100%

[github-actions]

Benchmark for 62c90e0

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	424.7±23.54µs	495.1±93.87µs	83%
Expression (Parser)	3.1±0.09µs	3.7±0.40µs	82%
Hello World (Execution)	424.2±15.39µs	491.9±19.30µs	84.00000000000001%
Hello World (Lexer)	1140.3±40.36ns	1188.6±48.79ns	96%
Hello World (Parser)	1353.0±50.97ns	1245.3±44.53ns	109.00000000000001%
Symbol Creation	469.4±28.16µs	612.3±38.99µs	70%
fibonacci (Execution)	4.6±0.10ms	5.4±0.14ms	82%
	undefined	undefined	100%

[Razican]

I guess that the lexer/parser for hello world show like this because of the build infrastructure, they are consistently better in my machine (at least the lexer). This is ready for a merge, in any case.

We should probably add Windows as a testing target, though, so that we can catch these faster.

[HalidOdat]

We should probably add Windows as a testing target, though, so that we can catch these faster.

I would like to add this. It shouldn't be hard with github actions.

[Razican]

I would like to add this. It shouldn't be hard with github actions.

Sure, let's do it in a different PR :) do you want to work on it?

[HalidOdat]

Sure, let's do it in a different PR :) do you want to work on it?

Yes. I'm really interested in github Actions.

[Razican]

This is ready for a merge :)

[jasonwilliams]

I’ll give it another test

[github-actions]

Benchmark for 5c1bdd6

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	355.0±11.97µs	446.1±26.96µs	74%
Expression (Lexer)	2.2±0.13µs	2.2±0.17µs	104%
Expression (Parser)	5.0±0.19µs	4.8±0.23µs	105%
Fibonacci (Execution)	4.0±0.15ms	5.1±0.12ms	74%
For loop (Execution)	377.6±13.69µs	535.1±15.93µs	58.00000000000001%
For loop (Lexer)	5.3±0.97µs	5.2±0.25µs	102%
For loop (Parser)	13.2±0.38µs	15.5±0.52µs	82%
Hello World (Lexer)	944.9±53.12ns	1012.6±40.26ns	93%
Hello World (Parser)	2.1±0.06µs	2.2±0.19µs	96%
Symbols (Execution)	386.9±13.68µs	552.4±29.46µs	57.00000000000001%
	undefined	undefined	100%

[github-actions]

Benchmark for d6394a6

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	310.8±16.26µs	368.8±11.75µs	81%
Expression (Lexer)	1898.1±59.73ns	1791.4±67.37ns	106%
Expression (Parser)	4.4±0.22µs	4.1±0.12µs	108%
Fibonacci (Execution)	3.3±0.13ms	4.1±0.13ms	74%
For loop (Execution)	314.5±13.06µs	452.9±13.82µs	56.00000000000001%
For loop (Lexer)	4.7±0.19µs	5.1±0.26µs	89.99999999999999%
For loop (Parser)	11.1±0.49µs	12.8±0.44µs	85.00000000000001%
Hello World (Lexer)	802.8±33.29ns	883.5±36.57ns	89.99999999999999%
Hello World (Parser)	1849.7±84.95ns	1884.1±101.27ns	98%
Symbols (Execution)	334.6±12.68µs	464.3±17.51µs	61.00000000000001%
	undefined	undefined	100%

[Razican]

I rebased, and it seems that the performance gains are very nice, especially on execution :)

[HalidOdat]

Looks perfect to me! 👍

[Razican]

Cool, let's wait for @jasonwilliams to finish the tests :)

[jasonwilliams]

works fine on Windows 10