feedback request, bug 3694 - ArrayBuffers, memmove, and signaling NaNs
(Meta: I'm assuming you'd prefer an on-list discussion rather than more comments in the bug?)
I think it's sensible to say that TypedArray.prototype.set doesn't require NaN canonicalization - or rather, that it forbids it and instead requires a bit-for-bit transfer.
I'm not sure I understand this:
I'm also inclined to put a statement somewhere in the TypedArray section that says that a bit level transfer must occur anytime an algorithm retrieves a value from a typed array using [[Get]] and immediately stores it back into a typed array, of the same type, using [[Set]]. It seem this would have to be stated as a mandatory requirement, in order to avoid interop hazards.
By "algorithm" are you referring to algorithms inside the spec, or more broadly any algorithm written in ES? If this is for algorithms in the spec, then I'm tempted to agree with this.
I've already been commenting in the bug. PTAL.
On Fri, Feb 13, 2015 at 5:13 PM, Filip Pizlo <fpizlo at apple.com> wrote:
I think it's sensible to say that TypedArray.prototype.set doesn't require NaN canonicalization - or rather, that it forbids it and instead requires a bit-for-bit transfer.
Why do you need to forbid NaN canonicalization rather than simply not requiring it?
Why do you need to require a bit-for-bit transfer instead of simply allowing it?
On Feb 13, 2015, at 5:13 PM, Filip Pizlo wrote:
By "algorithm" are you referring to algorithms inside the spec, or more broadly any algorithm written in ES? If this is for algorithms in the spec, then I'm tempted to agree with this.
I mean spec. algorithms
It'd be interested if anybody has any thoughts on my proposed solution for ecmascript#3694