[SR-2010] Implement SE-0113: Add integral rounding functions to FloatingPoint

[lattner]

Previous ID	SR-2010
Radar	None
Original Reporter	@lattner
Type	Task
Status	Closed
Resolution	Done

Additional Detail from JIRA

Votes	0
Component/s	Standard Library
Labels	Task, StarterBug
Assignee	@stephentyrone
Priority	Medium

md5: 24ed902defe2791a46ea5c4e09deaf97

Issue Description:

SE-0113 has been accepted, we need an implementation:
https://github.com/apple/swift-evolution/blob/master/proposals/0113-rounding-functions-on-floatingpoint.md

[lattner]

@stephentyrone do you plan to tackle this?

[stephentyrone]

I can make some time for it early next week. Providing default implementations will require the new integer stuff, and I'm not sure we have a good way to support Float80 without the importer being able to map long double, but we can add the API to FloatingPoint and wire up Float and Double quickly enough.

[stephentyrone]

Assigning this to myself so that it has an owner; if someone else wants to jump on it, however, feel free.

[lattner]

What do you mean that the importer can't map long double? Long double on X86 should come in as Float80.

[stephentyrone]

I agree that it should, but it doesn't seem to at present; there are some TODOs scattered in swift around this. I filed a radar yesterday when I became aware of the issue.

[stephentyrone]

Draft implementation (with horrible Float80 workaround) on Github. This is not an optimized implementation by any stretch of the imagination, but should get these working.

[stephentyrone]

#3443

[stephentyrone]

Merged in 9c08fbf.

[stephentyrone]

Optimization opportunity: https://bugs.swift.org/browse/SR-2045