if (<something> ! <something>) gives an if invalid format error, whereas all other symbols give an invalid operator error.

Also, !>, !>=, !<= and !< are not considered invalid operators. It seems the ! is ignored.

Thanks this has been fixed for the next version.

Similar thing :
if (<something> !% <something>) return * /if: '' unknown operator

//Var -s %a $iif(1 !%> 1,1,0) > * /if: '$true' unknown operator
">" is just an exemple, this bug with all char ( not tested )

if the var ( %> or else ) exist :
//Var -s %a 1 ,%b $iif(1 !%a 1,1,0) > * /if: '0' unknown operator

This is very strange.

the bug, in this case, is due to the usage of the % character, which, to my knowledge, is only used to start a variable name, or to return the remainder from a division calculation.

If this is so (or even if it's not), just what were you expecting the % sign to do as part of a comparator?

In your second example

which, substituting the value being assigned to %a (from the first part of the code) into the appropriate location in the second part, gives you
Since when is !1 (or not 1) a valid comparator between two different items (it's fine if being used as a full comparison in and of itself.)

In my two exemple, even if the variable name exist or not, it should return * /if: 'thing' unknown operator but it return '$true' or '0'.

does the variable is evaluated in the second exemple ? i think it should not be.

It would be evaluated, as only a single evaluation is required, which is the default when referencing a variable.

If you'd used $+(%,a) rather than %a then it would need the double evaluation to return the value the variable was set for, which, it appears, is what you were thinking of.

The bug is due to the fact that mIRC inconsistently reports errors, he's not suggesting the operator is a valid one. mIRC should report !%> unknown operator. There are two operands, and the operator doesn't start with a %, I think in this case it should not be evaluated.