search

ecraven / r7rs-benchmarks

Benchmarks for various Scheme implementations. Taken with kind permission from the Larceny project, based on the Gabriel and Gambit benchmarks.
271 stars 32 forks source link

Make Chez and Gambit safe by default. #15

Closed michaellenaghan closed 8 years ago

michaellenaghan commented 8 years ago

I think it would be ideal to benchmark both safe and unsafe code—but in the absence of multiple tests for the same implementation it seems to me that benchmarks should be safe by default.

michaellenaghan commented 8 years ago

I suspect that the -O6 optimization in Bigloo is also unsafe, but I can't find any clear documentation that confirms it. (-Obench is documented as unsafe, and I suspect -O6 = -Obench.)

michaellenaghan commented 8 years ago

I couldn't find absolute confirmation re: -O6 being unsafe in Bigloo, but I came close. If you agree with "safe by default" I'd suggest changing Bigloo from -O6 to -O3.

ecraven commented 8 years ago

Seems like a good idea, I'll change these and also change Bigloo to -O3

Thank you!