native, checker: builtin module for vlang

[Spydr06]

This PR is the beginning of making the builtin module work with the native backend.

As discussed on the Discord, native will utilize the already existing implementation of builtin used by the C backend, since libc is the official and most consistent API on many OSs apart from Linux.

Here is a list of things that still need to be implemented to get builtin working:

• enable the builtin module for native
• library linking support for Linux
• library linking support for MacOS
• library linking support for Windows
• global variables
• rebuild codegen of ast.AssignStmt to accept all expressions
• missing expressions, like ast.CastExpr (and many more)
• replace current inline implementations println(), assert(), exit(), C.syscall(), etc. by their builtin counterparts
• ...

(I will extend this list as needed)

For debugging purposes, it may be helpful if we could add a flag to toggle the inclusion of builtin in compile_native(), so that testing is still possible.

Please note:
Of course, this PR is far from being finished and still has a lot of errors. I'll be implementing all the missing stuff over time.

[Spydr06]

Expressions to implement:

• ast.CastExpr
• ast.Array (and others) support for ast.Ident
• flags field on arrays
• support for sizes > 8 in mov_deref()
• support for ast.UnsafeExpr, ast.CallExpr, ast.IfExpr, ast.MapInit, ast.InfixExpr, ast.ParExpr, ast.SelectorExpr, ast.PrefixExpr, ast.StringInterLiteral for ast.AssignStmt
• array support for <<
• ast.IndexExpr
• ast.CharLiteral
• ast.Nil
• expand expression support for ast.InfixExpr
• pointer support for != and ==
• support for pointer arithmetic
• ast.StringInterLiteral
• support more data types in ast.ArrayInit, also follow aliases!
• ast.ArrayInit outside of assignments
• in ast.InfixExpr support
• ast.ArrayDecompose
• support for exprs other than ast.Ident in ast.IndexExpr.right.left in ast.AssignStmt
• support for C.xxxx C identifiers -> checker error
• support for interlaced strings (ast.StringInterLiteral)
• expand expression support for ast.StructInit in ast.AssignStmt
• ast.HashStmt with other format than xx xx xx xx
• ...

[Spydr06]

Statements to implement:

• for-in
• multi-return for structs

[Spydr06]

Other bugs found

• ast.Alias type support
• cannot find variable if defined in unsafe block
• allocate_var() for sizes > 8
• allocate_var() for sizes < 4
• mov_reg_to_var() for all data types
• implement ast.ConstField
• rdx support in mul_reg; add this:
• same for mul_div
• support more expression of infix_expr.right in for_stmt()
• support le ge and ne in for_stmt infix_expr.left; add this:
• support more expressions in gen_amd64_exit()

[medvednikov]

A huge and important project.

Good luck!

[Spydr06]

thanks!
I think this could be the most significant change to make native usable, it'll take a lot of time though.

[Spydr06]

#17839 implements rune literals/printing, pointer comparison support and general helper function improvements

[Spydr06]

#18036 should make debugging much easier

[Spydr06]

Currently refactoring the codegen process: #18057

[Spydr06]

#18546 splits up the code generator into multiple files and refactors the AssignStmt codegen

[Spydr06]

#18703 implements support for ast.Alias types

[Avey777]

Are you still working

[Spydr06]

@Avey777
No, I no longer agree with internal decisions the leaders of this project have made and I have more important stuff to worry about at the moment.
I'll see if I get back to this sometime in the future.

[Avey777]

@Avey777

No, I no longer agree with internal decisions the leaders of this project have made and I have more important stuff to worry about at the moment.

I'll see if I get back to this sometime in the future.

We hope you will continue to come back, we have many friends in China looking forward to the 1.0 version of V language. V language syntax is really too good, can greatly reduce our development and view the code of mind! Native is important. 2 years have passed and a lot of things have changed. Maybe it's time for you to come back.

[Spydr06]

If native was so important I wouldn't have been the only one actively developing it. Also, this is not how open-source works. If people are not paid for their work they do it voluntarily in their free-time, and I have better things going on right now.

[JalonSolov]

Importance does not necessarily imply there is anyone else with the necessary skills, who is also willing to do the work.

You are correct about how open source works - contribute what you want when you want, but always take care of yourself, first.