amforth-devel

[Amforth-devel] Marker in butterfly amforth 3.4

[Ian J. <ij...@sa...>]

While chasing my serial problem I noticed that marker is not working.

I'd assumed this was some kind of serial issue but discovered that  
marker depends on get-current and set-current.

The 3.4 butterfly application (hex files) don't include set-current.

Looking at the source it looks like it needs

.include "words/set-current.asm"

added?

I have not tried this yet as I've been struggling with Avra.  Avra  
1.0x assembles the project after adding a device in the device  
structure in device.c.  Avra 1.2.3 fails to assemble with the  
following errors.

Pass 1...
../../core/words/udslashmod.asm(21) : Warning : Found CR (0x0d)  
without LF (0x0a). Please add a LF.
../../core/words/dnegate.asm(17) : Warning : Found CR (0x0d) without  
LF (0x0a). Please add a LF.
Error: Overlapping Code-segments :
   Start = 0x001A, End = 0x001A, Length = 0x0001
   Start = 0x001A, End = 0x001B, Length = 0x0002
Please check your .ORG directives !
Error: Overlapping Code-segments :
   Start = 0x001C, End = 0x001C, Length = 0x0001
   Start = 0x001C, End = 0x001D, Length = 0x0002
Please check your .ORG directives !
gmake: *** [bf.hex] Error 1

I gather around Avra 1.2.2 there was some overlapping segments  
checking added.

Should I look back a few revisions of amforth for the butterfly?

Thanks

Ian

Re: [Amforth-devel] Marker in butterfly amforth 3.4

[Erich W. <ew....@na...>]

Hello Ian,

Ian Jefferson wrote:
> While chasing my serial problem I noticed that marker is not working.
> 
> I'd assumed this was some kind of serial issue but discovered that  
> marker depends on get-current and set-current.
> 
> The 3.4 butterfly application (hex files) don't include set-current.
> 
> Looking at the source it looks like it needs
> 
> .include "words/set-current.asm"

You can do this in dict_appl.inc
...
.include "words/set-current.asm" ; for marker
.include "words/set-order.asm"   ; ./.

Then build a new amforth, which can load marker.
There may be better solutions ...

> I have not tried this yet as I've been struggling with Avra.  Avra  
> 1.0x assembles the project after adding a device in the device  
> structure in device.c.  Avra 1.2.3 fails to assemble with the  
> following errors.
> 
> Pass 1...
> ../../core/words/udslashmod.asm(21) : Warning : Found CR (0x0d)  
> without LF (0x0a). Please add a LF.
> ../../core/words/dnegate.asm(17) : Warning : Found CR (0x0d) without  
> LF (0x0a). Please add a LF.
> Error: Overlapping Code-segments :
>    Start = 0x001A, End = 0x001A, Length = 0x0001
>    Start = 0x001A, End = 0x001B, Length = 0x0002
> Please check your .ORG directives !
> Error: Overlapping Code-segments :
>    Start = 0x001C, End = 0x001C, Length = 0x0001
>    Start = 0x001C, End = 0x001D, Length = 0x0002
> Please check your .ORG directives !
> gmake: *** [bf.hex] Error 1
> 
> I gather around Avra 1.2.2 there was some overlapping segments  
> checking added.
I'm using avra 1.2.2 Build 94 (11. May 2007) (from 
avra_1.2.2-1_i386.deb) for my stuff.

However, there are some files with version 2 assembler 
around, e.g. those directly out of Atmels AVRStudio files, 
which avra cannot handle correctly. This may be another 
source of problems.

> 
> Should I look back a few revisions of amforth for the butterfly?
Always worth a try, but no garantees ;-)
I'm mostly working with atmega32 not atmega169 as on the 
butterfly.

Good luck!
Erich

Re: [Amforth-devel] Marker in butterfly amforth 3.4

[Matthias T. <mt...@we...>]

Ian,


> I'd assumed this was some kind of serial issue but discovered that  
> marker depends on get-current and set-current.
> 
> The 3.4 butterfly application (hex files) don't include set-current.
> 
> Looking at the source it looks like it needs
> 
> .include "words/set-current.asm"
> 
> added?

A fine place would be the appl_dict.inc file, the problem you describe
is really a big one: Which words should always be included and which
are used only seldom. I decided (for now) that marker is  a nice tool
but....

[avra errors]

> following errors.

Since I use the wine&atmel avrasm2.exe pair, I did not notice the
problems here, sorry for that. I just made a few changes to the
code and checked them into the subversion repository that address
both your problems. I could not test them, since "my" avra does
not know the atmega169.

> Should I look back a few revisions of amforth for the butterfly?

Not really, since I do not introduce those errors by intention. Earlier
versions _may_ not have the errors described above, but _do_ have other
errors (usually documented in the changelog on the website)

Bye
Matthias

Re: [Amforth-devel] Marker in butterfly amforth 3.4

[Ian J. <ij...@sa...>]

Hi Matthias,

Thanks very much.  I'm very excited about amforth.  It's a fantastic  
tool for these small platforms.

The butterfly seems to be one of the best platforms for such a  
system.  Hopefully just using amforth and posting will be helpful to  
someone.

I'm beginning to understand the compromises to such language platform.

On 29-Apr-09, at 4:36 AM, Matthias Trute wrote:

>> following errors.
>
> Since I use the wine&atmel avrasm2.exe pair, I did not notice the
> problems here, sorry for that. I just made a few changes to the
> code and checked them into the subversion repository that address
> both your problems. I could not test them, since "my" avra does
> not know the atmega169.
>

What Avra version are you running?
Do you have a butterfly?

I've patched Avra 1.0.x version (worked) and the 1.2.3 version  
(generated the overlapping Code-segment errors).  I don't really know  
what I'm doing but the patch seems simple and straightforward.  I can  
double check it and post it here.  It's basically one line in device.c  
but I think device.c is different depending on release.  The atmega169  
still has the old style .inc files available in Studio that seem to  
work with both tavarasm and avra.

I'll stick with the current releases.  I have not had this much fun in  
a long time.

Ian

Re: [Amforth-devel] Marker in butterfly amforth 3.4

[Ian J. <ij...@sa...>]

Marker problem solved!

My patched version of Avra 1.2.3 works fine with these amforth patches  
and adding set-current to appl_dict.inc. (blush)

Marker loaded just fine using the python shell program.

I made a minor change to bf.asm and perhaps a makefile change.  I will  
not be able to get back to this for a couple of days but I'll try to  
post the complete set of changes.  I should double check a few things.

Everything seems more stable now.  Perhaps amforth and I are getting  
to know each other.

Ian

On 29-Apr-09, at 4:36 AM, Matthias Trute wrote:

> Since I use the wine&atmel avrasm2.exe pair, I did not notice the
> problems here, sorry for that. I just made a few changes to the
> code and checked them into the subversion repository that address
> both your problems. I could not test them, since "my" avra does
> not know the atmega169.