Thread: [Amforth] .S “dot-s”

amforth is an extendible interpreter on microcontrollers

Brought to you by: erwaelde, mtrute

amforth-devel

[Amforth] .S “dot-s”

From: Enoch <ix...@ho...> - 2013-03-02 05:23:15

Hello Matthias & all,

Regarding .S

Indeed, says Forth 2012 RC1, "the format of the display is
implementation-dependent".

However, doesn't GFORTH display order make a better sense:

1 2 3 .s <3> 1 2 3  ok

While ours:

> 1 2 3 .s
3 2 1  ok

Thanks, Enoch.

Re: [Amforth] .S “dot-s”

From: Matthias T. <mt...@we...> - 2013-03-02 08:56:06

Hi Enoch,

> Regarding .S
> 
> Indeed, says Forth 2012 RC1, "the format of the display is
> implementation-dependent".
> 
> However, doesn't GFORTH display order make a better sense:

In which way: by adding the stack depth or the ordering?

> 
> 1 2 3 .s <3> 1 2 3  ok
> 
> While ours:
> 
>> 1 2 3 .s
> 3 2 1  ok

I agree that amforth's display looks confusing at the
first glance. Expecting a but? yes: I read from left to right
(as probably many others do too) and I expect the TOS to read
first.

Gforth's display confuses me. I always have to scan the entire
line to get the information I want.

Basically the built-in .s is a debugging aid for the
core development. I use it rather often when I dig in
internals. A more user friendly .s is better implemented
as forth code, DEPTH and PICK are available (I think they
are useful for that task).

Matthias

Re: [Amforth] .S “dot-s”

From: Enoch <ix...@ho...> - 2013-03-02 15:03:37

Matthias Trute <mt...@we...> writes:

> Hi Enoch,
>
>> Regarding .S
>> 
>> Indeed, says Forth 2012 RC1, "the format of the display is
>> implementation-dependent".
>> 
>> However, doesn't GFORTH display order make a better sense:
>
> In which way: by adding the stack depth or the ordering?
>
>> 
>> 1 2 3 .s <3> 1 2 3  ok
>> 
>> While ours:
>> 
>>> 1 2 3 .s
>> 3 2 1  ok
>
> I agree that amforth's display looks confusing at the
> first glance. Expecting a but? yes: I read from left to right
> (as probably many others do too) and I expect the TOS to read
> first.
>
> Gforth's display confuses me. I always have to scan the entire
> line to get the information I want.
>
> Basically the built-in .s is a debugging aid for the
> core development. I use it rather often when I dig in
> internals. A more user friendly .s is better implemented
> as forth code, DEPTH and PICK are available (I think they
> are useful for that task).

Hello Matthias,

IMHO .s should follow Leo Brodie's "Stack Picture" comment convention
(Thinking Forth pg. 151) as we are so accustomed to it. As developers we
can reverse .s order in our mind but it is an unnecessary burden on our
slow brain (called "System 2" in Daniel Khaneman's book, Thinking, Fast
and Slow) which is supposed to be focusing on bugs.

And if I have your ear :-)
I prefer (gforth): -1 -2 -3 .s <3> -1 -2 -3  ok
over (amforth): -1 -2 -3 .s 65533 65534 65535  ok
i.e., the regular numbers' format.

Thanks, Enoch.

>
> Matthias
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb

Re: [Amforth] .S “dot-s”

From: Matthias T. <mt...@we...> - 2013-03-02 18:12:35

Hi Enoch,

> And if I have your ear :-)

You'll (almost) always have it. ;)

> I prefer (gforth): -1 -2 -3 .s <3> -1 -2 -3  ok
> over (amforth): -1 -2 -3 .s 65533 65534 65535  ok
> i.e., the regular numbers' format.

Thats fine. Just remove the line in your application
dict_appl.inc file and add your version of .S

The reason why I use the u. instead of the "standard" .
is simple: Just switch to HEX. There you have
the same numbers that can be grepped in the LST and
MAP files from the assembler and match the numbers
in the DUMP output (mostly XT's and addresses).
As I said: .s is a debugging aid for _my_ core
development. I do not expect that it is unmodified
useful for other purposes.

Should I mention, that I collect good tools and ideas? ;)

Matthias

Re: [Amforth] .S “dot-s”

From: Enoch <ix...@ho...> - 2013-03-02 20:28:24

Matthias Trute <mt...@we...> writes:

> Hi Enoch,
>
>> And if I have your ear :-)
>
> You'll (almost) always have it. ;)
>
>> I prefer (gforth): -1 -2 -3 .s <3> -1 -2 -3  ok
>> over (amforth): -1 -2 -3 .s 65533 65534 65535  ok
>> i.e., the regular numbers' format.
>
> Thats fine. Just remove the line in your application
> dict_appl.inc file and add your version of .S
>
> The reason why I use the u. instead of the "standard" .
> is simple: Just switch to HEX. There you have
> the same numbers that can be grepped in the LST and
> MAP files from the assembler and match the numbers
> in the DUMP output (mostly XT's and addresses).
> As I said: .s is a debugging aid for _my_ core
> development. I do not expect that it is unmodified
> useful for other purposes.
>
> Should I mention, that I collect good tools and ideas? ;)

As recommended, removed words/dot-s.asm from dict_appl.inc and added:

: .s  ( -- )				\ stack picture listing order
    depth
    begin dup while dup pick . 1- repeat
    drop
;

-1 -2 -3 .s
-1 -2 -3  ok

I hope that I'm not nagging -- got more challenges for you if you don't
object ;-)

Regards, Enoch.


>
> Matthias
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb

Re: [Amforth] .S “dot-s”

From: Michael K. <mi-...@t-...> - 2013-03-03 00:10:11

Hi.

Am 02.03.2013 um 21:27 schrieb Enoch:

> : .s  ( -- )				\ stack picture listing order
>     depth
>     begin dup while dup pick . 1- repeat
>     drop
> ;

Maybe you want to have that in assembler. mk


-------------- next part --------------

Re: [Amforth] .S “dot-s”

From: Matthias T. <mt...@we...> - 2013-03-03 12:37:49

Enoch,

> I hope that I'm not nagging -- got more challenges for you if you don't
> object ;-)

You're welcome :)

Matthias

Re: [Amforth] .S “dot-s”

From: Michael K. <mi-...@t-...> - 2013-03-03 00:00:40

Hi.
..
> Should I mention, that I collect good tools and ideas? ;)
>>

Here is the old form for your collection. It is "Stack UnderFlow"  
save, telling "SUF" if you do one.
Maybe you have it already. Michael


-------------- next part --------------

Re: [Amforth] .S “dot-s”

From: Enoch <ix...@ho...> - 2013-03-03 01:01:32

"Michael Kalus" <mi-...@t-...>
writes:

> Hi.
>
> Am 02.03.2013 um 21:27 schrieb Enoch:
>
>> : .s  ( -- )				\ stack picture listing order
>>     depth
>>     begin dup while dup pick . 1- repeat
>>     drop
>> ;
>
> Maybe you want to have that in assembler. mk

Hello Michael,

I believe that we should have things coded in asm only when there is a
distinct advantage to it. 

During development I annotate my code with numerous "stack pictures"
(aka program verification invariants) and some .s printouts
(tracepoints) to confirm. I didn't understand why I have to read those
printouts in reverse, but not any more :-)

Regards, Enoch.

>
>
> -------------- next part --------------
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb

Re: [Amforth] .S “dot-s”

From: Erich W. <ew....@na...> - 2013-03-03 12:29:27

Michael,

On 03/03/2013 01:00 AM, Michael Kalus wrote:

> Here is the old form for your collection. It is "Stack UnderFlow"
> save, telling "SUF" if you do one.
> Maybe you have it already. Michael
>
>
> -------------- next part --------------

If you, too, want me to look at your code, you better
add it to the message text. At least I do not see
an attachment.

Erich

Re: [Amforth] .S “dot-s”

From: Michael K. <mi-...@t-...> - 2013-03-03 16:32:52

Oh, sorry. Forgot that the list strips attachments.
Michael


; Enochs .S
VE_DOTS:
     .dw $FF02
     .db ".s"
     .dw VE_HEAD
     .set VE_HEAD = VE_DOTS
XT_DOTS:
     .dw DO_COLON
PFA_DOTS:
; ( -- )
; stack picture listing order
     .dw XT_DEPTH
PFA_DOTS0:
     .dw XT_DUP
     .dw XT_DOCONDBRANCH
     .dw PFA_DOTS1
     .dw XT_DUP
     .dw XT_PICK
     .dw XT_DOT
     .dw XT_1MINUS
     .dw XT_DOBRANCH
     .dw PFA_DOTS0
PFA_DOTS1:
     .dw XT_DROP
     .dw XT_EXIT




; old form giving index#, address and value of stack items.
; ( -- ) Tools
; R( -- )
; stack dump
VE_DOTSS:
     .dw $ff02
     .db ".s"
     .dw VE_HEAD
     .set VE_HEAD = VE_DOTSS
XT_DOTSS:
     .dw DO_COLON
PFA_DOTSS:
     .dw XT_CR
     .dw XT_DEPTH, XT_1PLUS
     .dw XT_GREATERZERO
     .dw XT_DOCONDBRANCH
     .dw PFA_DOTSS0
; <old.s>
     .dw XT_SP_FETCH
     .dw XT_DEPTH
     .dw XT_1MINUS
     .dw XT_ZERO
     .dw XT_DOQDO
     .dw PFA_DOTSS2
PFA_DOTSS1:
     .dw XT_SLITERAL
     .dw 2
     .db "[]"
     .dw XT_ITYPE
     .dw XT_DUP
     .dw XT_I
     .dw XT_DUP
     .dw XT_UDOT     ; index of cell
     .dw XT_2STAR
     .dw XT_PLUS
     .dw XT_DUP
     .dw XT_DOLITERAL
     .dw 4
     .dw XT_UZERODOTR     ; address of cell
     .dw XT_DOLITERAL
     .dw 0x3A
     .dw XT_EMIT, XT_SPACE
     .dw XT_FETCH
     .dw XT_UDOT     ; content of cell
     .dw XT_CR
     .dw XT_DOLOOP
     .dw PFA_DOTSS1
PFA_DOTSS2:
     .dw XT_DROP
     .dw XT_EXIT
; </old.s>
PFA_DOTSS0:
     .dw XT_SLITERAL
     .dw 3
     .db "suf",0
     .dw XT_ITYPE
     .dw XT_EXIT



Am 03.03.2013 um 13:17 schrieb Erich Waelde:

> Michael,
>
> On 03/03/2013 01:00 AM, Michael Kalus wrote:
>
>> Here is the old form for your collection. It is "Stack UnderFlow"
>> save, telling "SUF" if you do one.
>> Maybe you have it already. Michael
>>
>>
>> -------------- next part --------------
>
> If you, too, want me to look at your code, you better
> add it to the message text. At least I do not see
> an attachment.
>
> Erich
>
> ---------------------------------------------------------------------- 
> --------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb
> _______________________________________________
> Amforth-devel mailing list for http://amforth.sf.net/
> Amf...@li...
> https://lists.sourceforge.net/lists/listinfo/amforth-devel

[Amforth] Variations on .S (was: .S “dot-s”)

From: Erich W. <ew....@na...> - 2013-03-03 21:26:49

Hi,

On 03/02/2013 06:22 AM, Enoch wrote:
> Hello Matthias&  all,
> 
> Regarding .S
> 
> Indeed, says Forth 2012 RC1, "the format of the display is
> implementation-dependent".
> 
> However, doesn't GFORTH display order make a better sense:
> 
> 1 2 3 .s<3>  1 2 3  ok
> 
> While ours:
> 
>> 1 2 3 .s
> 3 2 1  ok
> 

One of the great things on Forth imho is that you can make
the system do, what your eyes please. So a few variations
on " .s " :

\ variations on dot-s
\ dot-s, one way, signed output:
: ds  sp@ sp0 1 cells - do i @  . -2 +loop ;

\ dot-s, one way, unsigned output:
: uds sp@ sp0 1 cells - do i @ u. -2 +loop ;

\ dot-s, the other way (reverse?), signed output:
: rs  sp@ sp0      swap do i @  .  2 +loop ;

\ dot-s, the other way, unsigned output:
: urs sp@ sp0      swap do i @ u.  2 +loop ;

\ dot-s, verbose, as it used to be in earlier versions of amforth:
:  dsv  depth dup 0 do i u. dup i - cells sp0 swap - dup u. @  . cr loop ;
: udsv  depth dup 0 do i u. dup i - cells sp0 swap - dup u. @ u. cr loop ;

And watch the show:

	|> .s
	|65530 5 4 3 2 1  ok
	|> ds
	|1 2 3 4 5 -6  ok
	|> uds
	|1 2 3 4 5 65530  ok
	|> hex uds
	|1 2 3 4 5 FFFA  ok
	|> rs
	|-6 5 4 3 2 1  ok
	|> urs
	|65530 5 4 3 2 1  ok
	|> dsv
	|0 1539 -6
	|1 1541 5
	|2 1543 4
	|3 1545 3
	|4 1547 2
	|5 1549 1
	| ok
	|>

Just never assume that your preferred solution solves the same
problem for everyone else. I read words/depth.asm for enlightenment,
and I chose the above without using pick, because pick needs to be
added to the dictionary. Obviously there is more than one way to
do this. You may *want* error checking. I don't. If I do a mistake,
well I excuse the system for crashing.

Cheers,
Erich

Re: [Amforth] Variations on .S

From: Enoch <ix...@ho...> - 2013-03-04 01:53:47

Hello Erich & all,

The focus of my .s comment was on Matthias non standard stack listing
order, not on the implementation. This decision, in my opinion, slows
down a Forth trained programmer who is accustomed to maintaining the
standard "stack picture" in his mind.

If you wonder what is "System 1" / "System 2" that I referred to in my
communication see http://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow
for a brief description.

Regards, Enoch. 

Erich Waelde <ew....@na...> writes:

> Hi,
>
> On 03/02/2013 06:22 AM, Enoch wrote:
>> Hello Matthias&  all,
>> 
>> Regarding .S
>> 
>> Indeed, says Forth 2012 RC1, "the format of the display is
>> implementation-dependent".
>> 
>> However, doesn't GFORTH display order make a better sense:
>> 
>> 1 2 3 .s<3>  1 2 3  ok
>> 
>> While ours:
>> 
>>> 1 2 3 .s
>> 3 2 1  ok
>> 
>
> One of the great things on Forth imho is that you can make
> the system do, what your eyes please. So a few variations
> on " .s " :
>
>
> \ variations on dot-s
> \ dot-s, one way, signed output:
> : ds  sp@ sp0 1 cells - do i @  . -2 +loop ;
>
> \ dot-s, one way, unsigned output:
> : uds sp@ sp0 1 cells - do i @ u. -2 +loop ;
>
> \ dot-s, the other way (reverse?), signed output:
> : rs  sp@ sp0      swap do i @  .  2 +loop ;
>
> \ dot-s, the other way, unsigned output:
> : urs sp@ sp0      swap do i @ u.  2 +loop ;
>
> \ dot-s, verbose, as it used to be in earlier versions of amforth:
> :  dsv  depth dup 0 do i u. dup i - cells sp0 swap - dup u. @  . cr loop ;
> : udsv  depth dup 0 do i u. dup i - cells sp0 swap - dup u. @ u. cr loop ;
>
> And watch the show:
>
> 	|> .s
> 	|65530 5 4 3 2 1  ok
> 	|> ds
> 	|1 2 3 4 5 -6  ok
> 	|> uds
> 	|1 2 3 4 5 65530  ok
> 	|> hex uds
> 	|1 2 3 4 5 FFFA  ok
> 	|> rs
> 	|-6 5 4 3 2 1  ok
> 	|> urs
> 	|65530 5 4 3 2 1  ok
> 	|> dsv
> 	|0 1539 -6
> 	|1 1541 5
> 	|2 1543 4
> 	|3 1545 3
> 	|4 1547 2
> 	|5 1549 1
> 	| ok
> 	|>
>
> Just never assume that your preferred solution solves the same
> problem for everyone else. I read words/depth.asm for enlightenment,
> and I chose the above without using pick, because pick needs to be
> added to the dictionary. Obviously there is more than one way to
> do this. You may *want* error checking. I don't. If I do a mistake,
> well I excuse the system for crashing.
>
> Cheers,
> Erich
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb

Re: [Amforth] Variations on .S

From: Matthias T. <mt...@we...> - 2013-03-04 17:52:44

Hi,

> Hello Erich & all,
> 
> The focus of my .s comment was on Matthias non standard stack listing
> order,

Since there is no standard format, my variant cannot be non-standard.
Thats impossible by definition.

I updated http://amforth.sourceforge.net/TG/recipes/Dumps.html
with Erich's examples. There should everyone find a solution
that suits him/her.

Matthias

Re: [Amforth] Variations on .S

From: Enoch <ix...@ho...> - 2013-03-04 19:04:53

Matthias Trute <mt...@we...> writes:

> Hi,
>
>> Hello Erich & all,
>> 
>> The focus of my .s comment was on Matthias non standard stack listing
>> order,
>
> Since there is no standard format, my variant cannot be non-standard.
> Thats impossible by definition.
>
> I updated http://amforth.sourceforge.net/TG/recipes/Dumps.html
> with Erich's examples. There should everyone find a solution
> that suits him/her.

I like your approach, holding up to your convictions while respecting
other people opinions.

Regards, Enoch.

>
> Matthias
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb

Re: [Amforth] Variations on .S

From: Michael K. <mi-...@t-...> - 2013-03-05 18:47:51

Hi.
Lets take a look at some common Forths, Revision appended.
They do it in the typed order left to right.
Only VFX does it top down linewise, like amforth in its older days.
Though not a "standard", typed order is "Normative Kraft des  
Faktischen".
(google: "normative power of the factual")

Brodies I like best. Unfortunately he does not tell us how the  
negative numbers are treated.

Regards, Michael
:-)

-
Revision of Examples

gforth:
11 -22 33 .s <3> 11 -22 33  ok


win32forth:
11 -22 33 .s [3] 11 -22 33  ok...
(with a dot for each item, none if empty stack)


VFX Forth for Windows IA32:
11 -22 33 .s
DATA STACK
      top
              33 0000:0021
             -22 FFFF:FFEA
              11 0000:000B
  ok-3


SwiftX MSP430 EVALUATION 3.5.9 12-Apr-2011:
11 -22 33 .s
11 -22 33 <-Top  ok


And last but not least, the Masters Voice:
Starting Forth by Leo Brodie - Chapter 2
http://www.forth.com/starting-forth/sf2/sf2.html

<snip>
A Handy Hint
A Non-destructive Stack Print

Beginners who are just learning to manipulate numbers on the stack in  
useful ways very often find themselves typing a series of dots to see  
what's on the stack after their manipulations. The problem with dots,  
though, is that they don't leave the numbers on the stack for future  
manipulation.

The Forth word .S prints out all the values that happen to be on the  
stack "non-destructively"; that is, without removing them. Let's test  
it, first with nothing on the stack:

.S <0>  ok

As you can see, in this version of .S, we see at least one number.  
This is the number of items actually on the stack.

Now let's try with numbers on the stack:

1 2 3 .S <3> 1 2 3  ok

ROT .S <3> 2 3 1  ok
</snip>

Re: [Amforth] Variations on .S

From: Enoch <ix...@ho...> - 2013-03-05 19:16:55

Hello Michael,

Let's add one important case to your fine collection.
Here is .s implementation in openfirmware/forth/kernel/kernel.fth:

: (.s        (s -- )
   depth 0 ?do  depth i - 1- pick n.  loop
;
: .s         (s -- )
   depth 0<
   if   ." Stack Underflow "  sp0 @ sp!
   else depth
        if (.s else ." Empty " then
   then
;

Using the usual "stack picture" (of-course).  Since only Matthias has
commit rights, it's his call. I don't suggest to fork over this issue
though :-)

Regards, Enoch.


"Michael Kalus" <mi-...@t-...>
writes:

> Hi.
> Lets take a look at some common Forths, Revision appended.
> They do it in the typed order left to right.
> Only VFX does it top down linewise, like amforth in its older days.
> Though not a "standard", typed order is "Normative Kraft des  
> Faktischen".
> (google: "normative power of the factual")
>
> Brodies I like best. Unfortunately he does not tell us how the  
> negative numbers are treated.
>
> Regards, Michael
> :-)
>
> -
> Revision of Examples
>
> gforth:
> 11 -22 33 .s <3> 11 -22 33  ok
>
>
> win32forth:
> 11 -22 33 .s [3] 11 -22 33  ok...
> (with a dot for each item, none if empty stack)
>
>
> VFX Forth for Windows IA32:
> 11 -22 33 .s
> DATA STACK
>       top
>               33 0000:0021
>              -22 FFFF:FFEA
>               11 0000:000B
>   ok-3
>
>
> SwiftX MSP430 EVALUATION 3.5.9 12-Apr-2011:
> 11 -22 33 .s
> 11 -22 33 <-Top  ok
>
>
> And last but not least, the Masters Voice:
> Starting Forth by Leo Brodie - Chapter 2
> http://www.forth.com/starting-forth/sf2/sf2.html
>
> <snip>
> A Handy Hint
> A Non-destructive Stack Print
>
> Beginners who are just learning to manipulate numbers on the stack in  
> useful ways very often find themselves typing a series of dots to see  
> what's on the stack after their manipulations. The problem with dots,  
> though, is that they don't leave the numbers on the stack for future  
> manipulation.
>
> The Forth word .S prints out all the values that happen to be on the  
> stack "non-destructively"; that is, without removing them. Let's test  
> it, first with nothing on the stack:
>
> .S <0>  ok
>
> As you can see, in this version of .S, we see at least one number.  
> This is the number of items actually on the stack.
>
> Now let's try with numbers on the stack:
>
> 1 2 3 .S <3> 1 2 3  ok
>
> ROT .S <3> 2 3 1  ok
> </snip>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb