[fasmg] Protection against duplicated includes

Index > Macroinstructions > [fasmg] Protection against duplicated includes

Author

Thread

Jessé

Joined: 03 May 2025
Posts: 92
Location: Brazil

Jessé 04 Jan 2026, 01:46

There's one question I still have not figured out how to deal with, it is the classical duplicate include protection (a.k.a. 'include guard'), that issues an error when the same file (for any reason) gets included more than once.

In C, it seems like (I'm not good in C, so, don't judge me Laughing

Code:

#ifndef __SOME_UNIQUE_NAME__
#define __SOME_UNIQUE_NAME__
// do stuff
#else
#error Duplicated inclusion of <file>
#endif

If I try to do something similar using fasmg language, it ends in a "recursion loop" (I suppose), and doesn't end being compiled:

Code:

if ~ defined __F_INCLUDE_FILE_INCLUDED__
        define __F_INCLUDE_FILE_INCLUDED__
else
        err "Duplicated inclusion of file ", __FILE__, 10
end if

Quote:

flat assembler version g.l04h
Error: could not generate code within the allowed number of passes.
[ble: exit 2]

However, by doing the following, it actually works:

Code:

if ~ defined __F_INCLUDE_FILE_INCLUDED__
        define __F_INCLUDE_FILE_INCLUDED__
else
        err "Duplicated inclusion of file ", __FILE__, 10
end if

__F_INCLUDE_FILE_INCLUDED__ = 0   ; disabling this line makes fasmg go brrrrrrrrr...

Quote:

flat assembler version g.l04h
incldef.asm [3] f_include.inc [4]
Custom error: Duplicated inclusion of file f_include.inc
.

I know in anticipation that fasmg kind of look ahead (by its several passes), and this probably have to do with the previous model not working.
But I can't understand why the second one works.

So, I ask: is there a better fasmg way of doing such a thing?

Thanks in advance,

_________________
jesse6

04 Jan 2026, 01:46

bitRAKE

Joined: 21 Jul 2003
Posts: 4399
Location: vpcmpistri

bitRAKE 04 Jan 2026, 07:21

This might work for your use-case:

Code:

; We can make a global "#pragma once" mechanism if we assume all files have
; a unique path. Non-cannonical paths are fine as long as it's always the same
; form. Be mindful of file organization if you're going to use this.
; Primary benefits:
;       + files are not loaded multiple times
;       + no change to file needed (i.e. if/endif wrapper)

calminstruction onetime?: line*&
        match name any?,line
        jyes go
        err 'filename expected'
done:   exit

        local val,var
go:     compute val,+name               ; convert string to number
        arrange var,=ONETIME.val        ; unique variable
        check definite var
        jyes done

        compute val,0
        publish var:,val                ; constant

        arrange line,=include line
        assemble line
end calminstruction

Note: the consumer needs to know - which is a different policy than the product establishing the condition. If fasmg's __FILE__ always produced a canonicalized path then an ideal "#pragma once" would be possible. Maybe it should be called "module" to follow the C++ language changes?

_________________
¯\(°_o)/¯ AI may [not] have aided with the above reply.

04 Jan 2026, 07:21

bitRAKE

Joined: 21 Jul 2003
Posts: 4399
Location: vpcmpistri

bitRAKE 04 Jan 2026, 08:42

Jessé wrote:

But I can't understand why the second one works.

The trick is that forward references only work for "constants" (i.e. is only defined/set once). So, by setting the value more than once it becomes a "variable" and forward referencing is disabled for it.

_________________
¯\(°_o)/¯ AI may [not] have aided with the above reply.

04 Jan 2026, 08:42

Jessé

Joined: 03 May 2025
Posts: 92
Location: Brazil

Jessé 04 Jan 2026, 12:37

Nice, I did other tests and now it become clear.
Simply doubling the define statement inside if block also work (because it "pushes" another definition to the symbol, making it a variable definition, I guess).
Which means my idea was not wrong, it just missing understanding how it works.

The calminstruction example looks very interesting.
I'll save it to test it further.

04 Jan 2026, 12:37

Jessé

Joined: 03 May 2025
Posts: 92
Location: Brazil

Jessé 04 Mar 2026, 00:55

According to what I read (and understood) from fasmg manual, this should work:

Code:

if ~ definite __THIS_FILE__ ; This checks before
    define __THIS_FILE__ 0  ; This comes after
    do_stuff:
else
    err "Duplicated inclusion of this file.", 10
end if

But, it ends in the same way as if 'defined' was used.

Can anyone explain to me how 'defined' and 'definite' differs?
This is a very tough question to me, one that I really don't have a proper answer to myself.

In tempo: I officially adopt this:

Code:

if ~ defined __THIS_FILE__ 
    define __THIS_FILE__  
    define __THIS_FILE__
    do_stuff:
else
    err "Duplicated inclusion of this file.", 10
end if

Thanks in advance,

04 Mar 2026, 00:55

Tomasz Grysztar

Joined: 16 Jun 2003
Posts: 8515
Location: Kraków, Poland

Tomasz Grysztar 04 Mar 2026, 07:56

Jessé wrote:

Can anyone explain to me how 'defined' and 'definite' differs?
This is a very tough question to me, one that I really don't have a proper answer to myself.

The main problem here is that you use symbolic variables (defined with EQU or DEFINE). In classic fasm these were preprocessing-time variables, so they were evaluated (by replacing them with their text) before any actual assembly happened. To be at least partially compatible, fasmg performs a similar text substitution when evaluating expressions.

This means that if you do something like:

Code:

define __THIS_FILE__ 0 
if defined __THIS_FILE__

the expression to evaluate becomes simply:

Code:

if defined 0

Because "0" is considered a defined value, the effect is what you expected, but likely not for the reason you though it would.

This made it impossible for you to see how DEFINITE differs from DEFINED, because you ended up evaluating:

Code:

if definite 0

instead of checking an actual variable.

And this could get even worse. Consider a case of defining a variable with value being a name of a symbol that is not defined:

Code:

define FOO bar
if defined FOO

Here the condition is false, because it becomes:

Code:

if defined bar

Therefore to see the actual difference between the operators you need to switch to regular variables:

Code:

if defined FOO
        display 'defined'
end if

if definite FOO
        display 'definite'
end if

FOO = 4

If you use a symbolic variable, you may effectively create a symbolic link to another variable, hence the next example gives the same result (because the conditions check the state of the "bar" symbol):

Code:

if defined FOO
        display 'defined'
end if

if definite FOO
        display 'definite'
end if

define FOO bar
bar = 4

And here the result is different (because, again, it is the state of "bar" that is checked, not "FOO", and "FOO" is forward-referenced):

Code:

bar = 4

if defined FOO
        display 'defined'
end if

if definite FOO
        display 'definite'
end if

define FOO bar

See also the thread where we've been discussing the issue, and which lead to development of new features that made it possible to write a IFDEF macro that checks for the presence of a symbolic definition. This macro became one of the examples in final chapter of the manual.

04 Mar 2026, 07:56

Jessé

Joined: 03 May 2025
Posts: 92
Location: Brazil

Jessé 06 Mar 2026, 00:55

I think I got it now.
Many thanks for your effort explaing it to me.

I did a quick test that just proved it is correct, as well as my (from now) understanding:

Code:

if defined FOO
        display '0: defined elsewhere' bappend 10
else
        display '0: symbol not defined' bappend 10
end if

if ~ definite FOO
        ; FOO = 4
        if definite FOO
                display '1: definite now' bappend 10
        end if
else
        display '1: already definite' bappend 10
end if

if ~ definite FOO
        FOO = 4
        if definite FOO
                display '2: definite now' bappend 10
        end if
else
        display '2: already definite' bappend 10
end if

if ~ definite FOO
        FOO = 4
        if definite FOO
                display '3: definite now' bappend 10
        end if
else
        display '3: already definite' bappend 10
end if

It's output:

Quote:

artix-avlt:[jesse6]:~/ASM64/fasm2_latest/tests/defin(it)e$ fasm2 test1.asm
flat assembler version g.l2fx
0: defined elsewhere
2: definite now
3: already definite

2 passes, 0 bytes.

This, as I understand now, could be used as an effective include guard, using the "if skeleton" from the example above.

Please let me know if it has any potential problems, as I see you and many other experienced person in fasmg language suggesting somewhat different approaches.

Also, as a suggestion, I think your explanation here is good quality one, maybe you should include it somehow in the fasmg manual (I apologize in anticipation if it's already there, and I just missed it).
It clarified to me a lot regarding these aforementioned concepts.

06 Mar 2026, 00:55

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum