Patrick R. Michaud wrote: > On Mon, Nov 17, 2008 at 12:53:47PM -0800, Moritz Lenz wrote: >> >> Latest rakudo (r32779) dies on t/spec/S03-operators/increment.t: >> >> $ TEST_JOBS=1 make t/spec/S03-operators/increment.t >> t/spec/S03-operators/increment.rakudo.... >> 1..41 >> ok 1 - var incremented after post-autoincrement # TODO detect integer >> overflow/underflow >> [...] > > The test also fails entirely if the system doesn't have a bigint > library available. I'd like to see us split the various overflow > tests out of increment.t and into their own files. > > I think I've now updated Rakudo so that BigInt acts like > Integer with respect to Perl 6 methods, but since I don't > currently have the bigint library installed on my system I > don't have an easy way to test this. :-)
"works on my machine" So what should we do now with the tests? Should I unfudge all passing TODO tests, and we state that we need a bignum lib for rakudo? Or should they be skipped? (if skipped, I'll leave the task of fudging to you, Patrick, because that's much easier to do when you know what's failing ;-) > Also, I suspect that we may have an issue if a subclass of Integer > automorphs to a BigInt, but I haven't had a chance to test this yet. The Perl 6 view is that every Int is unbounded, and the 32bit (or whatever) integer is just an optimization. I guess once HLL mapping is in place that can be done somehow. (Or at least I hope...) Cheers, Moritz -- Moritz Lenz http://perlgeek.de/ | http://perl-6.de/ | http://sudokugarden.de/
