Hi everyone
- I'm Marco Mastropaolo, a programmer who - for as a hobby - is trying to create a Lua interpreter in C# with the intent of an as much "clean room implementation" as feasible (that is, implementing my grammar, interpreter, VM etc. instead of plainly converting the standard ones). I know, weird hobby.
For this reason, I'm trying to analyze those little details of the standard implementations which are not specified in the documentation..
One thing I don't get is how Lua rounds numbers when performing operations on the standard library.
For example, on my machine:
> return select(2, 1, 2, 3, 4, 5, 6, 7)
2 3 4 5 6 7
> return select(2.2, 1, 2, 3, 4, 5, 6, 7)
2 3 4 5 6 7
> return select(2.8, 1, 2, 3, 4, 5, 6, 7)
3 4 5 6 7
Given this, I deduced a rounding to nearest integer (confirmed for negative numbers).
However a few cases are curious:
> return select(2.5, 1, 2, 3, 4, 5, 6, 7)
2 3 4 5 6 7
> return select(2.50001, 1, 2, 3, 4, 5, 6, 7)
2 3 4 5 6 7
> return select(2.5001, 1, 2, 3, 4, 5, 6, 7)
2 3 4 5 6 7
> return select(2.501, 1, 2, 3, 4, 5, 6, 7)
3 4 5 6 7
At this point, I fell into temptation of peeking at the source.. and the conversion should be truncating (they are just type casts) unless some non-portable tricks are played with the machine floating point state register.
