Re: [Flashforth-devel] For-next loops - the New Design

[Herman A. <exp...@vn...>]

Igor
I think, we do not understand each other. Maybe I was not clear enough.
The current for-next in FF is very good for general purpose, when there 
is no problem with missing loop, and the skip function of  0 for  is 
beneficial. So, it needn't change.

My hypothetical n mfor...mnext would be an optional loop type to solve 
the full 16bit problem. It would be as fast, as the original for-next. 
In this case the meaning of n is the maximal distance from the reference 
position, that you want to rich.  If you substitute the for-next to 
mfor-mnext in your example, you will get the next results:

1 memtest
65535

2 memtest
131071

n memtest
n*65536-1

Because it is a non-standad loop structure, I think use it from a 
loadable text file, as Craig said.

BR
Attila


2015.05.28. 20:52 keltezéssel, om1zz írta:
> Attila
> with properly working 0 FOR you can do for example:
>
> 2variable count   \  just a counter to see the number of loops
>
> : memtest ( 64k_blocks -- )
>          0. count 2!      \ reset the counter
>          \ the BIG memory access loop, max 4,294,967,296 bytes
>          for
>                  65535 \ this should be normally 0
>                  for
> 		count 2@ 1. d+ count 2!   \ here you do something with memory
> 		next
> 	next
>          cr count 2@ ud. ;
>
> 1 memtest
> 65535  ok<#,ram>
> 2 memtest
> 131070  ok<#,ram>
> 16 memtest
> 1048560  ok<#,ram>
>    ok<#,ram>
>    ok<#,ram>
> 64 memtest
> 4194240  ok<#,ram>
>
> Now the count does not match as you miss a count every loop, but this is how the stuff works.
> It could be done with do-loop (which works with zero), but I think it would be more elegant with FOR.
> Igor
> ______________________________________________________________
>> Od: Herman Attila <exp...@vn...>
>> Komu: <fla...@li...>
>> Datum: 28.05.2015 17:21
>> Předmět: Re: [Flashforth-devel] For-next loops - the New Design
>>
>> I think the best solution of the  n for ... next  cycle, witch run
>> n-times, including n=0, as it is in the FF now. This principle
>> incopatible to cover the for full 16-bit range.
>>
>> To rich the full 16bit range must be defined an other for-next like
>> cycle, with different approach. In this case the goal is not to makes n
>> count, but to cover a non-zero long range. The cycle must run n to 0,
>> including n and 0. The 'next' decrements after testing for zero. We need
>> only to find the appropriate word-names.
>> Because it is mostly usable for memory addressing, maybe mfor-mnext.
>> n mfor r@ u. mnext   --->  n ... 0
>> 0 mfor r@ u. mnext   --->  0
>> 5 mfor r@ u. mnext   --->  5 4 3 2 1 0
>> 65535 mfor r@ u. mnext   --->  65535 65534 ... 1 0
>>
>> BR
>> Attila
>>
>>
>> ------------------------------------------------------------------------------
>> _______________________________________________
>> Flashforth-devel mailing list
>> Fla...@li...
>> https://lists.sourceforge.net/lists/listinfo/flashforth-devel
>>