These improvements are a bit more complex than all others I've yet proposed/done.

You could add a "sar" operator which works like the ASM instruction. I've no idea how to do this although this should not be to difficult if somewhen Privalov would finish his "Guide to Fasm internals".

But more the "size reporting"-operand is the feature I currently miss the most.
I have some macros which should work differently depending on the size of an operand. Although you can compare registers like "AL" and "AX" directly to define the size and you can compare numbers by their value to get the right size,

there is no way (I think so) to define the size of a memory operand passed to a macro. "eqtype" does not go.

Any idea?

Should I open a poll?

There is a trick that allows to do it in macro, a bit complicated though, here a sample for 32-bit mode:

Also with similar trick you can whether the 32-bit or 16-bit mode is used and adjust the tests.

Yes, this works, but looks a bit provisory. Perhaps I should write something for myself when the "Guide to Fasm internals" gets finished.

Do you remeber, once, we fasm-people also used macros for alignments lol:

Privalov, what about rol and ror?

Rol and ror would be nice to, and not to difficult to implement. (GFI)

It is only a matter of concept: you could add lots and lots and lots of operators, but this will be a waste of time/code size, since you won't probably use all. So here are some usefull operators you probably won't use any time soon:

trn: reverses bit pattern
chksum: calculates bitwize checksum
ibs/xbs: insert/extract bit string
rcl/rcr: rotate with 2^n + 1 bits (9,17,33...)

... many many others.

I think, not every operator should be implemented directly, since this causes the code to become messy.

i.e. you would mess up rcl/rcr with rotations through carry
r2l/r2r gives you a hint maybe that its using 2^n (maybe rnl/rnr)