Hello,

I think i don't understand the processing order regarding fasmg.
Using fasmg hll54, this source does not compile for me :

Code:

include 'format/format.inc' format PE64 CONSOLE macro library p&         irp dll, p                 dd 0, 0, 0                 dd RVA dll#_name, RVA dll#_table         end irp         dd 0, 0, 0, 0, 0 end macro macro import dll, p& ; import functions         dll#_name db \     ; 'name.dll'                 `dll, '.DLL', 0         dll#_table:        ; table         irp name, p        ; import rvas                 if used name                         name dq RVA _#name                 end if         end irp         dq 0         irp name, p        ; import names                 if used name                         _#name dw 0      ; ordinal=0                         db `name, 0      ; 'name'                 end if   end irp end macro entry start section '.text' code readable executable start:         sub rsp, 7*8         mov rcx, 0         call [GetModuleHandleA]                  mov rcx, 0         call [ExitProcess] section '.idata' import data readable writeable         library KERNEL32         include 'kernel.inc'                  ;import KERNEL32, \         ;       ExitProcess,\         ;       GetModuleHandleA

The error i get is :

Code:

win64.asm [39]:         call [GetModuleHandleA] macro call [1] macro parse_jump_operand [11] macro parse_operand [27] macro parse_operand_value [21] macro parse_address [21]:         ns.address = +addr Processed: @dest.address = +GetModuleHandleA Error: symbol 'GetModuleHandleA' is undefined or out of scope.

If i comment the line "'kernel.inc'" and uncomment the import lines, it does compile and produces a valid application.
The kernel.inc contains one long import statement with all the imports for KERNEL32.DLL, nothing more.
So the include statement gets processed later than i expected i guess ?

Thanks in advance.

Can you attach your kernel.inc file so that we have a complete source to reproduce the problem?
Also, please try increasing the number of errors shown with the "-e" switch (like "-e100"), perhaps the real underlying error is somewhere else.

Hello Tomasz,

I'm sorry, it was a really stupid mistake, i overlooked something simple.
After using the -e100 switch it showed me it could not find kernel.inc, that's correct because it is called kernel32.inc, after correcting that everything compiles just fine.
The confusion i had was that the short error tells me the cause is a unknown symbol whilst the entire file to include was missing.

Thank you for responding so quickly.

Regards.

This is the reason why I implemented a multiple error collecting into fasmg. I originally chose to display only one error in fasm 1 (and fasmg does the same by default) to avoid "error avalanches" that happen when a single mistake causes tons of error in the rest of the source, burying the original error in a flood of messages. However, since forward-referencing is a central theme to fasm, it is not unusual that a mistake causes errors earlier in the source text and that the first error encountered is not the original error. The option to trace multiple errors is then very helpful.