I have come across a strange error in compiling a fasm program. If I include the word "select" as a code reference and call it in the code, the following output from fasm occurs.

flat assembler version 1.67.2
C:\COBOLF\Cplr.fsm [410]:
.end start
\Fasm\Include\win32ax.inc [158] .end [14]:
import_wsock32
\Fasm\Include\win32ax.inc [119] import_wsock32 [0]:
{ common macro import_#lib \{ import lib,functions \} }
\Fasm\Include\macro/import32.inc [45] import [22]:
label dd RVA _label
error: symbol already defined.
> Execution finished.

It is easily overcome by changing the reference, but is "select" some sort of reserved word?

select. Since by using win32ax.inc you're automatically including wsock32.inc you get the clash (note that this doesn't mean your application imports from wsock32 if no function is used).

I hope I understood what you meant, it would be better if you include at least a few lines of code to see the reference you mention.

I had this strange thing once. If any of the predefined macros from FASM header files will be used in a place of your own functions, labels, etc. it may produce the clash. Simply rename the clashing keyword - call it Select<something> to be more readable and avoid the clash.

LocoDelAssembly:
I'm afraid I don't understand your reply.
Here is a super-simple example of what I mean.

include '\Fasm\Include\win32ax.inc'
start:
call select
invoke ExitProcess,0
select:
ret

AsmGuru62:
Yes, I did rename the reference, but I don't know what the clash might have been.

"select" is a Win32 API name.

It is used in the Winsock API.

So when you include 'win32ax.inc' you also get 'wsock32.inc'.

And 'wsock32.inc' defines the label "select".

So when you try to define the label "select" there is a clash with the definition inside 'wsock32.inc'.

Thanks Revolution - that explains the situation.

btw, the worst thing is that FASM gives incorrect error message - something completely different than what is going on. It should simply say on the line where I define the function: "<name>: identifier already defined". Maybe Tomasz will see this post and fix it?..

AsmGuru62, the problem is that it is actually the includes that are redifining since the "offending line" in wsock32.inc doesn't gets assembled until the programmer uses the ".end entry_point" macro invocation which happens at the end of the source (i.e. after "select:ret").

Still, what you say is somewhat doable by using this inside the official includes: But this won't tell which line of your source is causing this problem so you'll still need to search.

Good point.

I have IDE which generates code, so I have all the function entries, locals - all generated, so I do not use the end macro. That is why the message was so cryptic.