I've been writing all the code using MPLAB 3.99. I did choose
relocatable objects so that I could reuse code fragment with
ease (perhaps even make librarys).

Table read has to be made with a jump/call table. (With chips
like 16x84 or 12c50x) This is simple when page boundries are
not crossed. With non-relocatable code it is simple in that
case as well. (Covered well in the Microchip AppNotes)

However, with *relocatable* code it seems impossible. Example
of code:

PROG  CODE      ; Relocatable code
        ....
if ( (HIGH TableStart) & 1 )
        bsf PCLATH,0          ; set PCLATH right eg. this way
endif
        ...
        ...     ; make the computed call here
        ...
TableStart
        retlw 1     ; table here
        retlw 2
        ...

This code makes MPLAB complain about the expression being
too complex. (at the if command) Very logical as address is not
know before linking. Yet, the address is needed at the time of
code assembly.

Unfortunately this problem does arise with every external label,
even if the external label was defined by putting some table at
*known* (fixed) location. Well, it is still external and therefore
not known. A multi-pass assembler would solve this one
with ease.

Anyone found a way around these problems? Perhaps some other
approach that might work? Table reading is very common thing
to need. I'd hate to have to write non-relocatable code just for
that!


--
Harri Suomalainen     mailto:haba@cc.hut.fi

We have phone numbers, why'd we need IP-numbers? - a person in a bus