Fix compile time set cardinality #7558
Conversation
| 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, | ||
| 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, | ||
| 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7 | ||
| ] |
There was a problem hiding this comment.
- how was this generated? (i'm assuming not by hand...)
- could this be generated via CTFE?
- if not generated by CTFE, at least needs a test that tries all int8 values
There was a problem hiding this comment.
I generated it with some quick & dirty code that I didn't want to keep around forever. I agree that it would be better to write some proper code and generate it at compile time, since it's now basically impossible to verify it by just reading it. However, it doesn't really make much of a difference. This table only needs to be verified once, because it will never ever change (but as a counter point, I have added hard-to-verify constants that are wrong before).
There was a problem hiding this comment.
ok how about at least adding the quick and dirty code in a comment in that function? better than nothing IMO
There was a problem hiding this comment.
This table only needs to be verified once, because it will never ever change
Unfortunately it wasn't verified at all afaict. :-) Please use compile-time evaluation for this.
The issue was that duplicates in set literals are never discarded, so just counting the elements gives the wrong result.
Fixes #7555