[Coco] Back to DEF FN

Tue Jan 28 14:08:58 EST 2014

If I'm reading the code right, the original value of the parameter 
variable is saved before the formula is evaluated. Then the original 
value is put back again after. So the value of P should be unchanged.

The flag for FN values only applies to the function name (say the X in 
DEF FNX(P) = P + 3).

BTW, the example code below has a syntax error - there's no function 
name following "FN".

On 14-01-28 11:35 AM, Aaron Wolfe wrote:
> There is a bit flag for each variable entry that indicates whether it
> is a DEF variable.  I believe this means scalar P in BASIC is not the
> same variable as P in the DEF function.  Somebody more familiar might
> know for sure.
>
> On Tue, Jan 28, 2014 at 1:29 PM, Arthur Flexser <flexser at fiu.edu> wrote:
>> Well, this raises the more general question of what happens if a dummy
>> argument used in a function definition has the same name as a scalar
>> variable used by the program.  For example:
>>
>> 10 DEF FN(P) = P+3
>>
>> 20 P=20
>>
>> 30 Z=FN(5)
>>
>> The value of Z should now be 8, but what is now the value of P?  Is it
>> still 20, or has it changed to 5?
>>
>> Art
>>
>>
>>
>> On Tue, Jan 28, 2014 at 1:14 PM, Aaron Wolfe <aawolfe at gmail.com> wrote:
>>
>>> There are two tables of variables in BASIC as William mentioned.  One
>>> holds scalars, one holds arrays.  The ARYDIS flag causes the variable
>>> lookup routine to simply skip the table of arrays when looking for a
>>> match to the input variable name.  In the DEF FN case, this
>>> effectively prevents arrays from ever being used as the variable
>>> argument.  In your example the DEF FN(B) would always refer to the
>>> scalar B, never the array B.
>>>
>>> On Tue, Jan 28, 2014 at 12:49 PM, Arthur Flexser <flexser at fiu.edu> wrote:
>>>> Aaron, what does that "array search disable" do?   I'm guessing that if
>>> you
>>>> define FN(B)=B+3, that the array search disable means it makes no
>>>> difference if the dummy variable B happens to be the name of a declared
>>>> array or not.  Is that correct?
>>>>
>>>> Art
>>>>
>>>>
>>>> On Tue, Jan 28, 2014 at 12:29 PM, Aaron Wolfe <aawolfe at gmail.com> wrote:
>>>>
>>>>> Extended Basic Unravelled contains a detailed disassembly of the DEF
>>>>> code.  Here is the relevant bit:
>>>>>
>>>>>
>>>>> 1015 8880 BD B2 6A // JSR LB26A
>>>>>      SYNTAX CHECK FOR '('
>>>>>
>>>>> 1016 8883 C6 80 // LDB #$80
>>>>>     GET THE FLAG TO INDICATE ARRAY VARIABLE SEARCH DISABLE
>>>>>
>>>>> 1017 8885 D7 08  // STB ARYDIS
>>>>>     AND SAVE IT IN THE ARRAY DISABLE FLAG
>>>>>
>>>>> 1018 8887 BD B3 57 // JSR LB357
>>>>>     GET VARIABLE DESCRIPTOR
>>>>>
>>>>> 1019 888A 8D 25 // BSR L88B1
>>>>>     'TM' ERROR IF STRING
>>>>>
>>>>> 1020 888C BD B2 67 // JSR LB267
>>>>>     SYNTAX CHECK FOR ')'
>>>>>
>>>>> As you can see, there is no loop or any other mechanism that might
>>>>> allow more than a single variable to be parsed.
>>>>> The docs might be inconsistent or confusing, but the code is clear.
>>>>>
>>>>> -Aaron
>>>>>
>>>>>
>>>>> On Mon, Jan 27, 2014 at 7:37 PM, Rogelio Perea <os9dude at gmail.com>
>>> wrote:
>>>>>> Just got delivered a library copy of Peter Vernon's "Making The Mos of
>>>>> Your
>>>>>> TRS-80 Color Computer", it was the bargain bin on Amazon
>>> (Prentice-Hall
>>>>>> Australia 1983 ISBN 0 7248 0752 7).
>>>>>>
>>>>>> Leafing through the pages I came to the section where each ECB
>>> command is
>>>>>> listed, an odd wording of the old DEF FN caught my eye:
>>>>>>
>>>>>> 10 DEF FN( A B C)=(A+B+C)/3
>>>>>> 20 INPUT"ENTER THREE NUMBERS";A,B,C
>>>>>> 30 D=FNA(A B C)
>>>>>> 40 PRINT"THE AVERAGE IS";D
>>>>>> 50 PRINT
>>>>>> 60 GOTO 20
>>>>>>
>>>>>> I was perplexed for a bit. Could it be that the proper syntax on the
>>>>> CoCo's
>>>>>> DEF FN requires the variables to be separated by a space instead of a
>>>>>> comma? could it be *that* simple?
>>>>>>
>>>>>> I retyped the MOD routine into the CoCo as:
>>>>>>
>>>>>> 10 DEF FNRE=(N1 N2)=N1-INT(N1/N2)*N2
>>>>>> 20 CLS
>>>>>> 30 INPUT"NUMBER 1";N1
>>>>>> 40 INPUT"NUMBER 2";N2
>>>>>> 50 PRINT
>>>>>> 60 PRINT N1;"MOD";N2;"IS";FNRE(N1 N2)
>>>>>> 70 PRINT
>>>>>> 80 GOTO 30
>>>>>>
>>>>>> And it worked. This routine above is based on one shown by Lewis
>>>>>> Rosenfelder in "Basic Faster And Better & Other Mysteries" book. I was
>>>>> on a
>>>>>> roll and ported another one from Rosenfelder's (date day # finder):
>>>>>>
>>>>>> 10 CLS
>>>>>> 20 DEF FNJD(Y M
>>>>>> D)=(M-1)*28+VAL(MID$("000303060811131619212426",(M-1)*2+1,2))-((M>2)
>>> AND
>>>>>> ((Y AND NOT -4)=0))+D
>>>>>> 30 INPUT"YEAR (1901-2099)";Y
>>>>>> 40 INPUT"MONTH (1-12)";M
>>>>>> 50 INPUT"DAY (1-31)";D
>>>>>> 60 PRINT
>>>>>> 70 PRINT"THAT IS THE";FNJD(Y M D);"DAY OF THE YEAR"
>>>>>> 80 PRINT
>>>>>> 90 END
>>>>>>
>>>>>> Still smiling as I type this. The CoCo ECB book sins in being sparse
>>> at
>>>>>> best on covering one of the most underrated functions in the BASIC
>>>>>> repertoire, one that can come useful if applied properly. It had been
>>>>> years
>>>>>> (decades actually) since the first time I fiddled with DEF FN and it
>>> was
>>>>>> disappointing back then that I could not get it to work with 2 or more
>>>>>> arguments... I was using the syntax I knew from the TRS-80 Model I and
>>>>> III
>>>>>> BASIC separating the arguments by commas.
>>>>>>
>>>>>> With all this, the CoCo's DEF FN is still limited to numeric
>>> functions as
>>>>>> far as I know; ran a routine trying to define a string variable
>>> function
>>>>>> (simple concatenation) and the CoCo returned a type mismatch error. Oh
>>>>>> well, having this found to work with multiple variable arguments is
>>> in my
>>>>>> eye *the* discovery of the 21st century. Old ECB CoCo style :-)
>>>>>>
>>>>>>
>>>>>> -- RP
>>>>>>
>>>>>> --
>>>>>> Coco mailing list
>>>>>> Coco at maltedmedia.com
>>>>>> http://five.pairlist.net/mailman/listinfo/coco
>>>>>
>>>>> --
>>>>> Coco mailing list
>>>>> Coco at maltedmedia.com
>>>>> http://five.pairlist.net/mailman/listinfo/coco
>>>>>
>>>>
>>>> --
>>>> Coco mailing list
>>>> Coco at maltedmedia.com
>>>> http://five.pairlist.net/mailman/listinfo/coco
>>>
>>> --
>>> Coco mailing list
>>> Coco at maltedmedia.com
>>> http://five.pairlist.net/mailman/listinfo/coco
>>>
>>
>> --
>> Coco mailing list
>> Coco at maltedmedia.com
>> http://five.pairlist.net/mailman/listinfo/coco
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco
>