Parse Floating Point Constants #2195
Conversation
… decimal representation of real numbers
jiahanxie353
force-pushed
the
frontend-parse-fp
branch
from
1c8e8c8
to
ce90f1d
Compare
| }, | ||
| "mem_write": { | ||
| "data": [ | ||
| 4.2 |
There was a problem hiding this comment.
Oh the second thought is that it only proves that it reads/writes the same value, but doesn't prove that it's not necessarily intepretting 4.2 as the decimal representation..
There was a problem hiding this comment.
Can you explain what you mean by that? The comment is lacking enough context to make sense to me.
| @@ -26,7 +27,8 @@ num_lit = ${ | |||
| ~ ( "d" ~ decimal | |||
| | "b" ~ binary | |||
| | "x" ~ hex | |||
| | "o" ~ octal ) | |||
| | "o" ~ octal | |||
| | "f" ~ float) | |||
There was a problem hiding this comment.
One challenging thing about this is that we are hardcoding IEEE-754 encoding for floating point and might eventually want to support things like posits (something that @Mark1626 was working on) so we might want a different way to specify floating point numbers?
@sampsyo thoughts? One thing I was thinking is that the syntax is only available in a magical primitive:
cf1 = ieee754_const(5.11231); // Error if exact representation is not possible
This PR tries to address #2061
Support: