Closed WChung43 closed 3 years ago
By the way, I discovered a bug in fl+
and fl-
when testing the SDL integration.
The movapd
instruction (>=128-bit copy) should be replaced by movsd
(64-bit copy). A fix can be found here:
By the way, I discovered a bug in
fl+
andfl-
when testing the SDL integration.The
movapd
instruction (>=128-bit copy) should be replaced bymovsd
(64-bit copy). A fix can be found here:2dd764b#diff-b6c8fa84f32f5d7e9930a605d234098840001abc94ed39344ae0bc0b4a5f2927R734-R752
Don't worry. I saw that bug when testing the read function and I fixed it as part of this pull request in the edits for compile.rkt.
Hello. This is my small update to my implementation of the Floating point numbers for #1. I have updated read by doing a minor edit and uncommenting, so that the read file can parse Flonums, and I have fixed the compiler and a86 to remove bugs for boxing the result of float arithmetic operations during compilation.
The summary of the changes I made are:
read.rkt
compile.rkt
fl+
andfl-
which would cause errors. Fixed this by replacing all instances ofMovapd
with the correct instruction ofMovsd
float.rkt
Possible things to add next: