APDL Public Domain 1/APDL_PD1A.iso/program/assembler/sasm/Docs/Updates

Text File | 1995-10-02 | 5.4 KB | 116 lines

                Additions and amendments from version 1.70
                ------------------------------------------

This file documents additions and amendments made after version 1.70. If
you have a version before 1.70 you will probably find that there have
been so many changes (mainly due to a 'proper' front end appearing) that
you are advised to read it all again. This file will draw attention to
any changes made after that version so that you won't need to read the
whol;e manual to find out what has changed.


        1.71    Aug 95
        --------------
#STRUC (structure) directive introduced. This has (almost) the same
syntax as #AREA exacept that it ends with #ES instead of #EA. The
difference is that it does NOT reserve any memory, it just sets up the
offsets. These can then be used to address affsets in any data area, eg.
in an extension of the Wimpslot, in a file header etc.

Because #STRUC does not reserve any memory it can't use the label on the
first line to point to it, as is done with #AREA, for example -

        # struc
                first   4
                second  &40
                third   28
                forth   4
        # es

Would set up the variables first, second, third and forth such that if a
register was set to point to a certain piece of memory then -

        <instruction>,[<reg>,#first]            points to <reg>+0
        <instruction>,[<reg>,#second]           points to <reg>+4
        <instruction>,[<reg>,#third]            points to <reg>+&44
        <instruction>,[<reg>,#forth]            points to <reg>+96

and the data area will take a total of 1000 bytes. If you want 'first'
to be offset from the start of the area you would put the value on the
first line after #STRUC. Note that all values must be numeric (in hex or
decimal), variables cannot be used. As you can see, this is particularly
useful directive, and has equivalents in many high level languages.


        1.73    Sept 95
        ---------------

HS and LO are now accepted as conditions with conditional ADRL, MOVL,
etc. This hadn't been done before because these were never really part
of the ARM instruction set, the 'real' conditions are CS and CC, the
others were tacked on when Basic IV became Basic V. As I learned to
write ARM code before the textbooks mentioned HS or LO I have never
used them and so never noticed that they weren't there.

SASM_ERR variable
-----------------
It was suggested by a user that it would be very useful if the SAsm
error count were made 'available' to other routines. Errors are now
counted in the variable SASM_ERR (I know it's upper case and FP, but
it's more obvious and there's less chance of a clash).

If, for example, you have a routine that would only need to be called if
there's benn no errors during assembly then you could check that
SASM_ERR is 0 before you call it. Similarly if you have 'internal' error
checks in your own routines and you want to make sure that the error is
counted 'officially' then you can increase the count (#SASM_ERR+=1).

This is particularly useful after the #END directive as it enables you
to write LOCAL error handlers in functions and still record the errors.
In the main part of the code you can simple use #ERROR <num>,<message>
and the SAsm error handler will intercept it.


        Still to do (soon, I hope!)
        ===========================

These are some of the things that are either under serious development
or that are on the 'to do' list and will be dealt with when I have time.
This list is not necessarily in any particular order, and so features
may appear at any time (or not at all!). please do write if you have a
particular need.

A major re-write has been going on under the surface for some time to
enable SAsm to refer all errors back to the Module which will then
decide how to report them (error file, printed to screen, pass on to
desktop, etc.). When this is complete SAsm will then have complete
control of its error handling and I will be able to intoroduce eroor
throwback, either throught the DDE module (which I want to avaoud
because I can't supply it with SAsm) or through my own system. At this
time I will probably abandon the Utility version os SAsm, and I may then
introduce a 'key file' registration file as with all my other Shareware
programs so that you can get the latest version of SAsm from any source.

Floating point instructions are also under consideration. I must confess
that I can't see why anyone would want to write machine code and then
screw it up with FP, but if you really must have them ......

The whole Macro Library system is being completely re-written. The
library creation tool will then be a desktop tool whch will make things
much easier. Multiple libraries will also be permitted. I have had a few
bug reports with the Library system, so please bear with me I don't want
to waste time on the current system at present.

There is no reason why SAsm can't have functions added to output code in
AOF. I have considered this in the past, but I hadn't bothered because
most C programmers won't use asssembler unless they're absolutely forced
to, and AASM is now bundled with Acorn C, so there didn't seem much
point (SAsm was really intended for 'stand alone' machine code
programs). However, three people so far have pointed out to me that
although Beebug C has an in-line assembler it's only really suitable for
very simple routines, and they would like to be able to use SAsm. So,
I'll look at it again and try to arrange something.

        David Holden - September 1995