GAS-like anonymous labels.

Is it possible to use gas-like anonymous labels in fasm?

They work similary to fasm anonymous labels, but you can define multiple instances of them... For example, look at the following code:

Code:

cmp ax, bx
je equal
;Do something (if different)
jmp @f
equal:  ;Do something
@@:
;Continue

It's basically a normal if-then-else structure... but in asm... And in fasm we are required to define at least one non-anonymous label, thus polluting the name space (which can be a problem if my function has many constructs like this one).

In gas, I would do:

Code:

cmpw %ax, %bx
je 1f
#Do something (if different)
jmp 2f
1:  #Do something
2:
#Continue

(The important thing here is not the AT&T syntax idiosyncrasies, but the anonymous label syntax)

As you can see, in gas we avoid defining a label when we have this kind of constructs. This can be important if we want to put this code in a macro or so, but that doesn't matter in the case... I don't like to define macros that subvert the language semantics... My problem is name-space pollution, which is sometimes difficult to manage.
There is also the problem of when I don't know what name to give to the label. Namely because the algorithm is excessively generic or self-explanatory. In nasm I always kept naming the labels as ".loop1", ".loop2", ".loop3", etc, and I think that's not very comfortable to just give systematic names to labels... I'm not a compiler!

I've read some parts of fasm manual, primarily the parts related with the preprocessor, but I found no ideal way to implement this in the code's side.
I thought I could write a macro to define a GAS-style anonymous label (taking the number as an argument), and then a macro to access the last, or the following, label defined with a given number.
The approach I thought about was defining a preprocessor variable with the name of the last systematic name created by the preprocessor. Then, the name of the following "version" of the label would be found through concatenation (because the preprocessor is unable to do calculations, remember?). The only problem I found is that we can't ask the preprocessor to behave differently whether a constant is defined or not. This is possible with the assembler, but not with the preprocessor.

Besides, I've thought about an alternate way to solve the above problem. What about a special symbol called "@ff", meaning the second anonymous label counting from here.

JJ