I'm attempting to move a project to FASMG, and I'm trying to achieve identical output as it was with current version of my project with the FASM1 to ensure that everything is ok with my new macros. So, I will definitely see any differences in the output of the standard x86 instructions.

The first difference I've encountered that "push -1" is not the same as "push 0xFFFFFFFF" anymore. Any particular reason or it wasn't planned?

No, this was not an intended change. I'm going to correct it.

I have uploaded the corrections. Please test them.

It seems that it works. Thank you. I'll continue porting of my project. If I'll find some other controversial differences, I'll write about them here.

Thank you. These macros have not yet been tested much, as very few people tried using them.

Another controversial difference: Also 5 bytes instead of 2 in the FASMG. 0.0 always equals to 0 in binary form. So, 2-byte instruction is possible here.

I think it is better to check how float looks like when it is treated as a dword. For example, 0x00000001 equals to 1.4E-45, so "push 1.4E-45" also can take just 2 bytes.

I was not sure if I wanted this feature in the basic set of macros, but after all it is not that complex in implementation, so I added it now.

Thank you very much. I think that it is better when default set of macros for x86 will generate as compact machine code as possible.