feat: add NewObject completion

[jaschdoc]

Closes #6326

[jaschdoc]

Was testing out the completers and got an error from the program below (which is not a valid program but I was midway through typing it).

pub def main(): Unit \ IO =
    println("The sum from 0..42 is")


pub def f(x: Option[a]): Option[a] = match

#
# An unexpected error has been detected by the Flix compiler:
#
#   head of empty list
#
# This is a bug in the Flix compiler. Please report it here:
#
# https://github.com/flix/flix/issues
#
# -- Flix Compiler --
#
# Flix Version : 0.46.0
#   incremental: All
#
# -- Java Virtual Machine --
#
# JVM Version  : 21.0.1 (2023-10-17)
# JVM Vendor   : GraalVM Community
# JAVA_HOME    : /home/_/.sdkman/candidates/java/21.0.1-graalce
# System       : Linux
#
# -- Stack Trace --
java.util.NoSuchElementException: head of empty list
	at scala.collection.immutable.Nil$.head(List.scala:662)
	at scala.collection.immutable.Nil$.head(List.scala:661)
	at ca.uwaterloo.flix.language.phase.TypeReconstruction$.visitExp(TypeReconstruction.scala:195)
	at ca.uwaterloo.flix.language.phase.TypeReconstruction$.visitDef(TypeReconstruction.scala:35)
	at ca.uwaterloo.flix.language.phase.Typer$.$anonfun$visitDef$1(Typer.scala:179)
	at ca.uwaterloo.flix.util.Validation$.mapN(Validation.scala:429)
	at ca.uwaterloo.flix.language.phase.Typer$.visitDef(Typer.scala:178)
	at ca.uwaterloo.flix.language.phase.Typer$.$anonfun$visitDefs$2(Typer.scala:162)
	at ca.uwaterloo.flix.util.ParOps$.$anonfun$parTraverseValues$1(ParOps.scala:106)
	at ca.uwaterloo.flix.util.ParOps$.$anonfun$parMap$2(ParOps.scala:64)
	at java.base/java.util.concurrent.ForkJoinTask$RunnableExecuteAction.exec(ForkJoinTask.java:1423)
	at java.base/java.util.concurrent.ForkJoinTask.doExec(ForkJoinTask.java:387)
	at java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(ForkJoinPool.java:1312)
	at java.base/java.util.concurrent.ForkJoinPool.scan(ForkJoinPool.java:1843)
	at java.base/java.util.concurrent.ForkJoinPool.runWorker(ForkJoinPool.java:1808)
	at java.base/java.util.concurrent.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:188)

[magnus-madsen]

Thanks-- can you open a ticket for the bug?

[magnus-madsen]

For the completer, it should fill in all methods you have to implement.

[magnus-madsen]

For example, if you write new java.util.Comparator then it should fill in the compare method you have to implement.

[jaschdoc]

For example, if you write new java.util.Comparator then it should fill in the compare method you have to implement.

Thanks! Working on it 😄 🏃 🔨
Just had to figure out what I was doing and what the goal was hehe

[magnus-madsen]

Goal is like Intellij IDEA when you get it to fill out everything you have to implement. So a bit like InstanceCompleter.

[jaschdoc]

Hmm it's kind of awkward to do completions as the new parser fails when writing new ##java.lang.Runnable or just new java.lang.Runnable so it seems the completion is not even run? I'm not sure if the parser has anything to do with the LSP side of things, though.

[jaschdoc]

Ok this is close to done. The only thing I'm not sure how to approach is that formatType doesn't prepend ## to native types. This means that NewObjectCompleter has to do it manually. However, Arrays pose a bit of a challenge, since we want formatType to format these but also add ## if the type argument is a native class. E.g., Array[Array[java.lang.Object]] should be Array[Array[##java.lang.Object]] (not adding regions).
Maybe @mlutze has a suggestion for a fix?

[magnus-madsen]

A few screenshots of different completions?

[mlutze]

Ok this is close to done. The only thing I'm not sure how to approach is that formatType doesn't prepend ## to native types. This means that NewObjectCompleter has to do it manually. However, Arrays pose a bit of a challenge, since we want formatType to format these but also add ## if the type argument is a native class. E.g., Array[Array[java.lang.Object]] should be Array[Array[##java.lang.Object]] (not adding regions). Maybe @mlutze has a suggestion for a fix?

I think our best option is probably to make a new formatter for generated code. We might be able to use the SimpleType interface.

[jaschdoc]

Ok this is close to done. The only thing I'm not sure how to approach is that formatType doesn't prepend ## to native types. This means that NewObjectCompleter has to do it manually. However, Arrays pose a bit of a challenge, since we want formatType to format these but also add ## if the type argument is a native class. E.g., Array[Array[java.lang.Object]] should be Array[Array[##java.lang.Object]] (not adding regions). Maybe @mlutze has a suggestion for a fix?

I think our best option is probably to make a new formatter for generated code. We might be able to use the SimpleType interface.

Roger that. I was afraid you were going to say that hehe :D

[magnus-madsen]

Before writing a lot of code, just reuse FormatType. Perfect is the enemy of good here. We don't want yet another formatter to maintain.

[jaschdoc]

[magnus-madsen]

General comments:

• We should also add support to Parser2. Maybe @herluf-ba can help?
• Add some doc comments for the helper functions.
• I don't think you need System.lineSeparator. I think \n is sufficient because it is interpreted by VSCode.