Perform incremental compilation using official compiler API

[fwcd]

Investigate support for incremental compilation using the official Kotlin compiler API instead of the mechanism currently used (as described in the README), because this might improve performance considerably.

[fwcd]

This might be related:

Caused by: java.lang.OutOfMemoryError: GC overhead limit exceeded
	at java.util.Arrays.copyOfRange(Arrays.java:3664)
	at java.lang.String.<init>(String.java:207)
	at java.lang.String.substring(String.java:1969)
	at java.lang.String.subSequence(String.java:2003)
	at kotlin.text.StringsKt__StringsKt.commonPrefixWith(Strings.kt:775)
	at kotlin.text.StringsKt__StringsKt.commonPrefixWith$default(Strings.kt:765)
	at org.javacs.kt.position.PositionKt.changedRegion(position.kt:95)
	at org.javacs.kt.CompiledFile.oldOffset(CompiledFile.kt:114)
	at org.javacs.kt.CompiledFile.parseAtPoint(CompiledFile.kt:72)
	at org.javacs.kt.signatureHelp.SignatureHelpKt.signatureHelpAt(signatureHelp.kt:22)
	at org.javacs.kt.KotlinTextDocumentService.signatureHelp(KotlinTextDocumentService.kt:128)
	at sun.reflect.GeneratedMethodAccessor76.invoke(Unknown Source)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:498)
	at org.eclipse.lsp4j.jsonrpc.services.GenericEndpoint.lambda$null$0(GenericEndpoint.java:61)
	at org.eclipse.lsp4j.jsonrpc.services.GenericEndpoint$$Lambda$35/698288148.apply(Unknown Source)

[apatrida]

Is incremental compilation needed for code completion or for what purpose? How does the kotlin-web-demo handle completion?

https://github.com/JetBrains/kotlin-web-demo/tree/master/versions/1.2.50/src/main/java/org/jetbrains/webdemo/kotlin/impl

[fwcd]

@apatrida Each file needs to be compiled to an abstract syntax tree by the Kotlin compiler frontend. Using the AST, the language server can provide document symbols, code completion, hover and more.

The naive solution to this problem would be to compile the entire file every time the user makes an edit, but that would be extremly slow (and make your editor feel laggy). To address this problem, the language server currently extracts the expression around the cursor, puts it into a small "fake" file and tells the compiler to compile the "fake" file. Still, it would probably be better to use official compiler APIs as mentioned in the issue description.

The Kotlin demo seems to be promising in terms of performance, I will take a closer look at it soon.

[Strum355]

I don't know well enough how the current method works (lots of methods disappearing into 3rd party packages etc), but would this solve the synchronization issue where e.g. a property is referenced in one "expression" but doesnt exist yet, is added at the relevant location, yet the error that its missing still persists?

on a side note, is there a gitter or some other more convenient place where people can discuss this project with you?

[fwcd]

@Strum355 There is a new Gitter channel for this project:

[xiaoyunwu]

@fwcd I am also interested in incremental compilation, and what is the newest pointers on this? Thanks.

[weilbith]

Sounds very promising to improve performance. Is it complicated to add this?