the Parser migration from fastparse to cats-parse

[lvitaly]

Relates to #631

[YulawOne]

I think it worth to make some performance comparison between fastparse and cats-parse solution

[timzaak]

If it's going on?

[ghostdogpr]

@timzaak I don't think there was much progress, in case you're interested to contribute ;)

[timzaak]

I tried, but cats-parse api is now very different from fastparse. Almostly need rewrite all the code. and need wait this issue: typelevel/cats-parse#186
I will try to pass the Parser test cases. https://github.com/timzaak/caliban/tree/cat-parse

Honestly, It's may be a good idea to wait fastparse support Scala 3.

[lvitaly]

@timzaak @ghostdogpr I rewrote Parser with cats-parse but stuck with whitespace implementation. I will commit the last changes soon.

[timzaak]

cats-parse 0.3.1 has changed a lot comparing with 0.1
you may update it firstly

[timzaak]

@lvitaly I changed a lot depends on your work. remove fastparse implemention of whitespace, and try to insert it in everywhere needed. So Sad.

[timzaak]

https://github.com/timzaak/caliban/blob/cat-parse/core/src/main/scala/caliban/parsing/Parser2.scala#L32 the whitespaces can work

[timzaak]

@lvitaly @ghostdogpr
Parser2.scala now can pass all the test cases( except invalid syntax, becase cats-parse error is different from fastparse)

The performance would improve by handle whitespaces and trackback more carefully.

PS:

1. string value implementation is from cats-parse JsonString.
2. I firstly use cats-parse, so it needs code review before created merge.

[timzaak]

ref: typelevel/cats-parse#188 (comment)

[timzaak]

I rewrite it, and it looks good to me, I don't know if it is good to replace fastparse, or just provide an alternative choice for GraphQL parse.

[ghostdogpr]

@timzaak one thing that would be interesting is to run the caliban benchmark project with fastparse vs cats-parse. If there is not much difference maybe we could migrate entirely. If cats-parse is much slower maybe we only adopt it temporarily inside a "scala3" branch.

[ghostdogpr]

Closing in favor of #850