Greg Shrago

[Grammar-Kit](https://github.com/JetBrains/Grammar-Kit) takes PEG grammar and produces PsiBuilder-based parser and PSI classes. Parboiled-to-BNF conversion should be pretty straightforward but lexer extraction (at least a simple whitespace/word tokenizer) will take some time...

Friendly bump. Were my changes of any use?

@incaseoftrouble thanks for response I've extracted CONCAT'd names, and tried to replace `Fast` suffix with something you've proposed. 1. For maps with primitive keys (e.g. `Int2IntOpenHashMap`) everything goes smooth. `fastutil`...

On the second thought, `mergeInt` and `computeInt` overloads look not that bad, consider: `mergeInt(1, 2, (int a, int b) -> a)` or `m.mergeInt(2, 2, (IntBinaryOperator)(a, b) -> a)`. Source-compatibility would...

@vigna Do I understand correctly that you are referring to the following solution? ``` import java.util.function.*; class Scratch { public static void func(ObjectIntMap map) { map.mergeInt(10, 20, (a, b) ->...

I've pushed changes for Int/Long/Double maps as we discussed above (had to fix BinaryOperator.drv a bit). We can use new type-specific operators for all other primitive maps.

> OK, I understand a bit better now the situation. The problem is that this has to be fixed first at the interface level, or open hash maps will have...

Something wrong with 4be250e92d8ce6eb0efcaa4545046067340aaa9b. MERGE def is deleted and a new method with MERGE name is added in `Map.drv`. I can't compile my PR because it uses MERGE=merge and I...

Compiles OK now. Are there any questions left for this PR?

> To bring this in line with the overall restyling, we need to avoid #ifdef. Do you have some specific code lines in mind? > Did you try whether no...