dsimcha:
> even though Map gets fully inlined when used alone.

This is what I too have found in past.


> I wonder if this is a weakness of compiler inlining heuristics in general, 
> though.

I have a bug report open for LLVM about this :-)


> is this just an ugly corner case in the imperfect heuristics
> compilers use to inline functions?

Those heuristics are often fragile.


> if so, now I understand why old-school
> programmers don't trust compilers.

Good :-)
On the other hand I have heard that the Scala compiler is better in inlining 
callables.

Bye,
bearophile

Reply via email to