Go Target rethink

[parrt]

Start a new conversation about the Go target here from #3718 (comment)

@jimidle has agreed to take a look

[parrt]

We should definitely think about the visitor pattern to see if it's correct:

#2504

I also remember either getting a crash or no error message upon syntax error when I tried Go target.

[parrt]

I think there were some data race fixes also for Go: #3137 so the multi threaded correctness might be OK. @Joakker would you care to comment on the current thread safety of Go?

[parrt]

Another thoughtful analysis about the various interfaces in Go target: #1843

[kaby76]

I recommend reading what I said in:

• Go target generates tables for parser and lexer using non-unique names #3452 where two different grammars are placed in one directory, e.g., grammars-v4/csharp/, and name collisions are created because parsers are created with global variable names but scoped by package name. It complicates the design of a program that uses two parsers. Suggest renaming global variables so that the recognizer name is included. I could work around it, but it's a hassle. Take the code in grammars-v4/csharp and just try to write a program that works and see what I mean. See also to Golang target @ member variables are global #3264. Everything is declared globally, scoped by package.
• Go target generated parser code does not have any calls to initialize base class parsers #3446 where it's impossible to call "base class" (in this design, really a "base struct") initialization of a recognizer. Every field in "base classes" have to use "zero initialize". Try porting grammars-v4/sql/plsql/ to Go.
• Go runtime missing export of "String()" #3443

It feels like the Go target wasn't really tested with the superClass option, trying the "target-agnostic format" model.

[KvanTTT]

It feels like the Go target wasn't really tested with the superClass option, trying the "target-agnostic format" model.

It's the right time to create new tests for superClass option.

[parrt]

oh wow. We don't have any? zoiks.

[parrt]

hmm... Global variables in Go target? yuck.

[parrt]

Very useful links, thanks @kaby76 Hopefully @jimidle can address these as well.

[parrt]

Looks like we can close as @jimidle has corrected most stuff.