Re: [Amforth-devel] [bug] missing "then" not caught by compiler

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

So I call it a feature then, ok.
Kalus Michael wrote:
> Hi Erich.
> 
> Its more like "keep it simple".
> 
> Tiny microprocessor forth systems usualy dont abort unstructured code.
> For example in fig forth or Rockwell AIM65 Forth the ;s just returned, 
> no error checking at all:
> 
> aim65:
> code: ;S  ( IP <-- RP )
> PLA
> STA 9F
> PLA
> STA A0
> JMP NEXT
> 
> fig forth entsprechend, zB. 8086:
> code: ;S
> MOV SI,[BP]  ( IP <-- R1 )
> INC BP
> INC BP
> JMP NEXT
> 
> 
> And amforth does not check for it too:
> : ';'  compile exit [ ;
> 
> 
> Big Forth systems do abort compilation if code ist unstructured. This is 
> done by setting local variables which are checked by ; finaly.
> 
> Use depth of stack as a minor check. If stack depth has changed after 
> compilation of a word look twice what was done there. So if you find a 
> bad definition, enclose it in DEPTH and if it does not read 0 1 like in
>  depth : test 1 2 3 . . . ; depth .s <2> 0 1  ok
> revise code.
> 
> 
> Michael
> 
> 
> 
> 
> 
> Am 31.03.2009 um 20:56 schrieb Erich Waelde:
> 
>> Hello,
>>
>> amforth, rev.736 (3.4)
>>
>>
>> amforth 3.4 ATmega32 ok
>> > : test1 0= if -1  ( no then ) ;
>>  ok
>> > : test2 0= if -1 else 0 ( no then ) ;
>>  ok
>> > 1 test1 .
>> --- hangs here until reset ---
>> amforth 3.4 ATmega32
>> > 1 test2 .
>>  ?? -13 8
>> >
>>
>> if statements missing the trailing "then" will compile
>> without the compiler barking. I consider this a bug.
>>
>> calling such a word will produce a "hang" in one case.
>> This can be viewed as another bug, but that is not
>> important. In the other case it is producing an
>> error message at least.
>>
>> Cheers,
>> Erich
>>
>