There was a problem hiding this comment.
LGTM; I took the liberty of fixing the sqrt test value. (The values in cmath_testcases.txt should all be the "ideal" values - i.e., those that would arise from a perfectly correctly-rounded implementation.)
Sorry, something went wrong.
|
Does it make sense to keep that after #102046? Let's leave as is - they're not causing any problems, so there's no reason to change. And who knows - we might encounter some situation in the future where we want this code to work on systems without signed zeros ... |
Sorry, something went wrong.
|
Let's leave as is - they're not causing any problems, so there's no reason to change. Then it probably does make sense
|
Sorry, something went wrong.
|
Then it probably does make sense [...] No, sorry, not worth it. We can worry about that if a relevant system ever shows up. |
Sorry, something went wrong.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.