vasm 2.0

[phx]

The vasm 2.0beta is approaching the end of its beta-test phase and I would like to release it before my holidays in July. The biggest new feature - support for cpu-architectures which don't use 8-bit bytes - is not relevant for Amiga, but there are also many changes which are.

So this is last opportunity to report any annoying bugs or important features which you would like to see in the release. If you have something, speak, or contact me by email!

[frakswe]

Minor thing but I'd like to see CC?=gcc set in the Makefile , then it picks up the env var if found else defaults to gcc.

[phx]

Does it really default to gcc, when CC is not defined your environment?

I tried it here (NetBSD with gcc, no CC environment set) and my "GNU Make 4.2.1" calls "cc" instead of "gcc". Which is not a problem for me, as cc is hard-linked to gcc. But I'm unsure if this would cause problems on other systems. Does Linux have a cc-link?

[frakswe]

Was set to "CC=gcc" on OpenBSD until recently , now it's "CC=cc" . If you feel uncertain then keep it as is, it's not that critical.

[Locutus]

Originally Posted by phx

The biggest new feature - support for cpu-architectures which don't use 8-bit bytes - is not relevant for Amiga, but there are also many changes which are.

Just curious, which architectures would that be? Vasm for pdp8 or something else?

pdp8 support would actually be nice, vasm is a bit nicer then PALBART.

[Locutus]

Ah, found the answer in the source code tarball.

[pipper]

@phx is it possible me to have an official source mirror repository on GitHub?

[copse]

It would be good to be able to use github's online tools :-)

[roondar]

Nice to hear v2.0 is approaching release. Looking forward to finding out what's new once it's released

[hitchhikr]

If something like that isn't implemented (yet) i'd like to have a specific command to be able to redefine a register alias several times (equr only allow to define a register alias once), like the equ/set couple (setr ?)

[phx]

Originally Posted by Locutus

Just curious, which architectures would that be? Vasm for pdp8 or something else?

A 16-bit and a 32-bit architecture for a start, but PDP8 is a nice idea! In the beginning our goal were byte-sizes which are a multiple of 8, but other sizes may indeed work now. I'm curious to try it. (The output to an 8-bit based file system will have some padding bits, of course).

Originally Posted by pipper

@phx is it possible me to have an official source mirror repository on GitHub?

What makes it "official"? The only official source can only be our private CVS server, which shows up as daily source snapshot every night. You are free to clone that to a public repository of course, as long as you don't modify it. But anybody could do.

Originally Posted by hitchhikr

If something like that isn't implemented (yet) i'd like to have a specific command to be able to redefine a register alias several times (equr only allow to define a register alias once), like the equ/set couple (setr ?)

Option

-regsymredef

should allow that since 9 years.

[pipper]

Originally Posted by phx

What makes it "official"? The only official source can only be our private CVS server, which shows up as daily source snapshot every night. You are free to clone that to a public repository of course, as long as you don't modify it.

The “me” in my original question was not supposed to be there. Sorry for the confusion. Im asking for an official place on GitHub to find the sources. There are some “mirror” repos for vasm on there (among them one I created), but it’d be tedious to keep them up to date. It would be great to have a mirror done by the actual devs for 100% ground truth. It should be possible to even setup an automated mechanism to pull things from cvs directly(?!)

[Tigerskunk]

Nice...

would it be possible to have REPTN being used in labels?

Like

REPT 20
mylabel_REPTN:
ENDR
?

[phx]

Originally Posted by pipper

It would be great to have a mirror done by the actual devs for 100% ground truth. It should be possible to even setup an automated mechanism to pull things from cvs directly(?!)

You mean, including the history logs? No. Please understand that vasm is an open source project, and we also take suggestions and patches, but it is not a public community project. All development is private between Volker and me.

So a daily source snapshot is the best you can get. Not sure how much you can automate that.

[phx]

Originally Posted by Tigerskunk

would it be possible to have REPTN being used in labels?

Text replacements are only performed in macro mode. Although this could be done with some effort, I'm not sure if it is really required.
Currently you have two possibilities to reach your goal.

1. By a macro which defines a unique global label:

mylab   macro
lab\@:
        endm

2. By defining a new isolated block of local labels with

inline

:

        rept    20
        inline
        moveq   #2,d0
.1:     dbf     d0,.1
        einline
        endr

[paraj]

Originally Posted by phx

Text replacements are only performed in macro mode.

Here's a pattern I've used a couple of times:

colfunc macro
colfunc\<col>:
.cnt set 0
        rept screend
        ifne (col&(1<<.cnt))
        bset.b  d1,.cnt*rowbytes(a0)
        else
        bclr.b  d1,.cnt*rowbytes(a0)
        endc
.cnt set .cnt+1
        endr
        rts
        endm

col set 0
        rept (1<<screend)
        colfunc
col set col+1
        endr

[phx]

Originally Posted by paraj

colfunc\<col>:

Yes, that's also option 1 (macro). But using

\<symbol>

instead of

\@

is a smart variation, in case you need specific label names.

[paraj]

Originally Posted by phx

Yes, that's also option 1 (macro). But using

\<symbol>

instead of

\@

is a smart variation, in case you need specific label names.

Yes, just supplementing your answer
The specific labels come in handy if followed by (from same code):

colfuncentry macro
        dc.w    colfunc\<col>-colfunctab
        endm

colfunctab:
col set 0
        rept (1<<screend)
        colfuncentry
col set col+1
        endr

A little bit cumbersome, but just to show some real world code, and perhaps inspire Tigerskunk. Something like that might be hard to express properly with a "REPT n" label. While the current implementation might be a bit cumbersome to use (needing indirection) it is flexible enough to allow this code (from https://eab.abime.net/showthread.php?t=112182):

merge macro
x set \1
y set \2
x1 set \1-1
y1 set \2-1
rx set 4-\1
ry set 8-\2
merge\<x>\<y>:
        cmp.w   d\<ry>,d\<rx>
        bls.b   .x\@
        move.w  d\<ry>,(a1)+
        bra     merge\<x>\<y1>
.x\@:
        move.w  d\<rx>,(a1)+
        bra     merge\<x1>\<y>
        endm

ycnt set 4
        rept 4
xcnt set 4
        rept 4
        merge xcnt,ycnt
xcnt set xcnt-1
        endr
ycnt set ycnt-1
        endr

Anyway, this is just to say, that the feature is not required, and a "half-baked" approach wouldn't help in more advanced cases anyway.

BTW looking forward to v2.0

[Tigerskunk]

No offense, but this looks overly complicated for the simple iterator on global labels that I need here.

Anyway, you do you of course...

[phx]

The std-syntax module provides the

.irp

directive for "iterating repeat". You can specifiy some values which will be assigned to a symbol:

        .irp    sym,10,20,30,40
lab\sym:
        .byte   \sym
        .endr

Maybe I can implement an

IRP

directive into mot-syntax, if this is useful. (Although I don't like to invent new directives, which make vasm incompatible with other assemblers.)