Ben Lerner
Ben Lerner
According to GitHub there are merge conflicts, so it's not yet merge-ready. Also, the latest commit was just yesterday, so again, I want to be sure this is _complete_ (as...
@ds26gte please fix at least these things, which I found just _skimming_ the code, not even deep reading it yet: - in jsnums-tests: the code does not need `arrayEquals`, since...
Please don't remove the unused HAC functions yet; if anything that should be a separate PR, but it's low priority at most. Regarding `toFixnum` -- _why_ does `errbacks` propagate through...
Obvious argument: () is misleading as it's not the signature of the function Obvious counterargument: should the test name include the names of the function arguments? In examplar situations, that...
Comment from our zoom discussion: #1807 fixes the case of `log(1e+VERYBIG)`, which produces a number that's roughly `VERYBIG`. Do we know whether we handle `log(1e-VERYBIG)` correctly, and make sure that...
> `print(num-log(1 / num-expt(9, num-expt(5, 7))))` takes a long while and reports “roughnum overflow" fwiw
@jpolitz @sorawee ping -- I suspect that homework is finished by now... ;-)
The code above is indeed broken in current CPO. Not sure what the fix is, offhand and only from my phone, though ...
Oh that's an interesting ergonomic issue, and I'm kinda impressed we haven't encountered it before... I _think_ we could handle that using the pattern Joe suggested above, though I'm slightly...
Could be. I don't like the thought of hard-coding such a brittle warning pattern, but it may be worth it...