About std/haxe/Int64 inlining

[soywiz]

All Int64 methods are inlined for targets that don't have a custom implementation. https://github.com/HaxeFoundation/haxe/blob/development/std/haxe/Int64.hx#L253

So for example when adding two longs all that body is inlined, and at least on javascript target that would impact on size and probably even it will have worse performance (because js engines tends to optimize better smaller methods and to inline internally when required).

Non single-line / or with several param references, public and inlined methods: add, sub, mul, compare, ucompare, neg, preIncrement, eq, neq, and, or, xor, shl, shr, ushr, toInt, ofInt

Is this the desired behaviour?

[Simn]

Some functions there indeed appear to be rather large for inlining. However, before we touch anything we need some profiling data.

[delahee]

hi :)

Heavy +1 for profiling data. Some engines are better than others but we still to care about baseline ones. As a general matter it's way better to come and ask for optimization with a bunch of data otherwise there are always possible "counter scenarios" and we won't be able to trakc your particular use case / differentiate with general need.

Thanks.

2016-04-24 10:09 GMT+02:00 Simon Krajewski notifications@github.com:

Some functions there indeed appear to be rather large for inlining. However, before we touch anything we need some profiling data.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/HaxeFoundation/haxe/issues/5153#issuecomment-213911034

David Elahee