Use of TCHAR in UNICODE build

[MichaelK]

Hello all. I've been searching for hours for a solution to this problem I'm having. I'm having trouble assembling a program making use of TCHAR. ML throws an error saying "constant value too large" when I try to use TCHAR in a Unicode build or even WORD instead of BYTE. Here is an example source:

Code Select Expand


.386
.model flat, stdcall
option casemap:none

include windows.inc
include user32.inc
include kernel32.inc
includelib user32.lib
includelib kernel32.lib

.DATA
szTitle TCHAR "ASM Application", 0
szMessage TCHAR "Hello, world.", 0

.CODE
start:
invoke MessageBox, 0, ADDR szMessage, ADDR szTitle, MB_OK OR MB_ICONINFORMATION
invoke ExitProcess, 0
end start

And here is the command and parameters I'm trying to build with:

Code Select Expand


ml /c /Cp /coff /nologo /W3 /D__UNICODE__ /D__NO_NOISE__ /I \masm32\include /Fo obj/main.o src/main.asm

Is there something I'm missing? That's just what I can't seem to figure out :(

[hutch--]

Michael,

have a look at the HLHELP.CHM help file in MASM32. Under the heading Unicode support it explains how it works.

[Zen]

MICHAEL K.,
Hi. This is one of those weird problems where the solution is not obvious.
As you know, WINDOWS defines a TCHAR as a single byte character, using the ASCII standard codes,...unless, UNICODE is defined,...then, a TCHAR is defined as a two-byte Unicode character.
Most assembly programmers just use the single-byte character codes  because we all speak English. I think the problem is (and HUTCH mentioned this in a thread monthes ago), that the MASM copmpiler only recognizes the single-byte definition (which doesn't conform to the WINDOWS data type definition).
The solution is simple: find the MASM definition file and alter (temporarily) the value of the TCHAR character, to the double-byte UNICODE type.

Added as an edit: (OOPS,...HUTCH got the answer ahead of me)

[dedndave]

i had this problem a while back
moreover, i wanted the LENGTHOF operator to return the number of characters, whether it was built for UNICODE or ANSI

qWord (forum member) gave me a nice solution that i have been using without problems, since

Code Select Expand

;###############################################################################################

;tchr macro by qWord

tchr    MACRO   lbl,args:VARARG
    IFDEF __UNICODE__
        UCSTR lbl,args
    ELSE
        lbl db args
    ENDIF
        ENDM

;###############################################################################################

        .DATA

tchr szNoFile,   'No File Specified',0
tchr szTooBig,   'File Too Large',0
tchr szNotFound, 'File Not Found',0
tchr szAlready,  'File Already Exists',0
tchr szInpFile,  'Input File:',13,10,0
tchr szOutFile,  'Output File:',13,10,0

;***********************************************************************************************

[MichaelK]

Thank you everyone for your replies. Though the docs do not state it clearly, my understanding is that you cannot use TCHAR to statically initialize a variable but only dynamically. Like so:

Code Select Expand


szBuffer TCHAR "Some text", 0   ; Error
szBuffer TCHAR 50 DUP(0)        ; OK


Also, when I try to use UCSTR like in the help file, ML throws a syntax error. Like this:

Code Select Expand


UCSTR szBuffer, "Some text", 0


Am I missing something like an include file? I'm using the same include files like the one in my original post.

[dedndave]

Code Select Expand

UCSTR szBuffer, "Some text", 0

that assembles ok, here
make sure you are in either the .DATA or .CODE section (.CONST will also work)  :P

[MichaelK]

Unfortunately, not on my system. The whole program is exactly like the original one I posted except the inclusion of the UCSTR line (in the data section), but ML generates the following error message:

Code Select Expand

error A2008: syntax error : UCSTR

[dedndave]

ok
you have to include more files   :P

in this case, specifically \masm32\macros\macros.asm

but - to make things simple, get rid of the entire preamble and

Code Select Expand

include \masm32\include\masm32rt.inc

that file takes care of casemap, processor, model, and a number of includes

[MichaelK]

Thanks. It works perfectly now 
Would you recommend editing masm32rt.inc to include qWord's code snippet so the use of TCHR when in Unicode mode or otherwise becomes transparent?

[dedndave]

no - i only edit the masm32 files when a bug is found
otherwise, you will not be compatible with other forum members
you post a program that you can assemble, but others cannot - lol

put it in a seperate file
what i quite often do is to create an INC file for the project - that would be a good place

[MichaelK]

Got it. So I've created a separate include file to take care of that. This is how it looks (I modified the original snippet to use qWord's UCCSTR macro to enable support for escape sequences too):

Code Select Expand


TCHR MACRO lbl, args :VARARG
IFDEF __UNICODE__
UCCSTR lbl, args
ELSE
lbl db args
ENDIF
ENDM


I got a feeling today is gonna be awesome for me. Good start! 

[dedndave]

Code Select Expand

TCCHR MACRO lbl, args :VARARG
IFDEF __UNICODE__
UCCSTR lbl, args
ELSE
lbl db args
ENDIF
ENDM

spot the difference ?   

[MichaelK]

Pardon me to resurrect this dead thread. But what's the difference between "TCHR" and "TCCHR". AFAIK, "TCHR" isn't defined anywhere.  :icon_confused:

[dedndave]

TCCHR must be the name of someone's macro

TCHR is the name used by qWord for his macro
i prefer this macro because it returns correct values when used with LENGTHOF and SIZEOF operators

Code Select Expand

TCHR    MACRO lbl,args:VARARG

    IFNDEF __UNICODE__
        lbl db args
    ELSE
        UCSTR lbl,args
    ENDIF

        ENDM

[dedndave]

declare the macro early in the source (it generates no code)

then, to use it....

Code Select Expand

    .DATA

TCHR  szString,"MyString",0

now, it generates the correct ANSI string if the __UNICODE__ symbol is not defined
or the correct UNICODE string if the __UNICODE__ symbol is defined