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

Should safe and unsafe results be consistently marked and/or separated? #11

Closed michaellenaghan closed 8 years ago

michaellenaghan commented 8 years ago

There are two entries for Chez, effectively safe and unsafe. The fact that other entries are unlabelled makes it easy to assume that they're all "safe"—but, as the Gambit prelude shows, some of them are actually "unsafe".

ecraven commented 8 years ago

There are two entries for Chez, effectively safe and unsafe. The fact that other entries are unlabelled makes it easy to assume that they're all "safe"—but, as the Gambit prelude shows, some of them are actually "unsafe".

Where do you see "safe" and "unsafe" for Chez? Chez has chez and petite (as far as I understand compiled and interpreted).

For Gambit, I'd ask anyone familiar with gambitc to tell me how to improve the prelude, I don't actually use gambit myself at all :-/

michaellenaghan commented 8 years ago

Hmmm, I was looking at https://www.nexoid.at/tmp/scheme-benchmark-r7rs.html thinking they were from you; now I realize nothing says they are. In any event, they show entries for optimizelevel-2 and optimizelevel-3. 3, which is what the project currently uses, is unsafe; 2 is the default.

I guess the original question still stands, though: do you want unsafe results? If so, I'd suggest having both safe and unsafe and labelling them separately.

I'll tell you what. I'll close this issue, and I'll try to put together a pull request tomorrow. Then you can decide whether or not you want the changes.