Announcement

**George Bleck** · 13 Jan 2023, 02:54 PM

You can't (natively) and poking around that way would NOT be advisable.

PEEK$(address, count) can easily do what you need but, ignoring WHAT you wrote for a second (very vague example), what are you trying to do?

I've implemented an example as a MACRO here

Code:

#COMPILER PBWIN
#COMPILE EXE
#DIM ALL

'********************************************************************************
MACRO GetSect(paramSection) = PEEK$(C+(paramSection*X),X)
'********************************************************************************


'********************************************************************************
FUNCTION PBMAIN() AS LONG
'********************************************************************************
  LOCAL lDebug AS LONG: TXT.WINDOW "Debug", 10,10,50,75 TO lDebug 'Thanks Dave!

  LOCAL X AS LONG
  X = 5

  LOCAL C AS DWORD
  LOCAL D AS STRING

  D = "....1....2....3....4....5"
  C = STRPTR(D)
  TXT.PRINT GetSect(0) & " " & GetSect(1) & " " & GetSect(2) & " " & GetSect(3) & " " & GetSect(4)

  TXT.PRINT "Press any key to exit (just like the old days!)": TXT.WAITKEY$: TXT.END
END FUNCTION

**Frank Rogers** · 13 Jan 2023, 03:49 PM

I'm playing around with a merge record sort routine. I have one routine to sort by PartNo, one to sort by Description, one to sort by QuantityOnHand, ... It works with a hard coded field, but if I want to sort by another field it requires another sort routine which is 99% the same. I'm trying to knock down many routines into one general routine.

I could set up an equate to each field and then a CASE SELECT to pick the field but that feels messy.

I need to pass the starting field location and field length to the routine.

With the UDT I can just @pString[RecNo] = @pString[RecNo +1] but the compare is what is hanging me up

**Frank Rogers** · 13 Jan 2023, 04:02 PM

I think the only way to do it is with the Offset to the field and the Length.

MSort(sInventory, lRecLength, lOffset, lLength)

In the calling routine I will have to compute the Offset to PartNo, Desc, QtyOnHand... and the Length of the field 20, 50, 12...

Thanks George

**Eric Pearson** · 13 Jan 2023, 04:25 PM

I think you want ARRAY SORT... CALL. It allows you to write a function that determines the sort order. Perfect for multi-field sorts of UDTs, you can even mix ascending/descending sorts, field by field. Or am I not understanding what you need to do?

**Dale Yarker** · 13 Jan 2023, 05:03 PM

From post 1:

LOCAL C AS STRING PTR * X '<-------- I want to set this at runtime

"X" is the maximum string length of a fixed length string. Just compile-time. If a shorter string is assigned it is padded with spaces. L = INSTR(@C, $SPC) - 1 will give length except when length = X.

But better (IMO) would be dynamic strings. STRPTR() and LEN() are definitely run-time.

Cheers,

**Frank Rogers** · 13 Jan 2023, 05:19 PM

Originally posted by Eric Pearson View Post

I think you want ARRAY SORT... CALL. It allows you to write a function that determines the sort order. Perfect for multi-field sorts of UDTs, you can even mix ascending/descending sorts, field by field. Or am I not understanding what you need to do?

Hi Eric,

Yes, I tried DIM I(1 to MaxRec) as InvtRec AT STRPTR(sInventory) and used ARRAY SORT ... Call. It still required a case tree with equates for the different fields but it did work. In my testing my MergeSort was quicker than ARRAY SORT... but just barely. I'm sure that last sentence will start a tangent war on sort speed.

**Frank Rogers** · 13 Jan 2023, 05:21 PM

Originally posted by Dale Yarker View Post

From post 1: "X" is the maximum string length of a fixed length string. Just compile-time. If a shorter string is assigned it is padded with spaces. L = INSTR(@C, $SPC) - 1 will give length except when length = X.

But better (IMO) would be dynamic strings. STRPTR() and LEN() are definitely run-time.

Cheers,

I see what you are getting at Dale. What I also needed was the size of the pointer to index through the buffer. @sInventory[RecNo].Desc By changing the RecNo it would advance which record I was looking at.

**Dale Yarker** · 13 Jan 2023, 05:30 PM

. . . size of the pointer

????? 4 bytes.

**Stuart McLachlan** · 13 Jan 2023, 05:56 PM

Deleted

**Frank Rogers** · 13 Jan 2023, 06:24 PM

Originally posted by Dale Yarker View Post

????? 4 bytes.

By size I mean the size of the UDT indexed by the pointer. Or what the Pointer is pointing to. The size of the record in my case. Yes, the pointer is 4bytes, but what it points to is the UDT size.

Code:

TYPE MyRecord
  PartNo AS STRING * 20
  Desc AS STRING * 50
  Quantity AS LONG
END TYPE

LOCAL pInventory as MyRecord Ptr
pInventory = STRPTR(Buffer)
@pInventory[RecNo].PartNo

By incrementing the RecNo above it advances one record at a time or 74 bytes in this example. I'm trying to change the 74 at run time. Probably not possible.

**Eric Pearson** · 13 Jan 2023, 06:43 PM

Deleted

**Dale Yarker** · 13 Jan 2023, 07:26 PM

SIZEOF

But you did the TYPE/END TYPE so why don't you know? Unless some alignment is used.

See SIZEOF Help.

I'm trying to change the 74 at run time. Probably not possible.

Correct.

Forget UDTs, Load entire record to STRING, then PARSE COUNT and PARSE$. After the two strings get next four bytes and use CVL(). (for example, the TYPE in post 11)

Cheers,

**Stuart McLachlan** · 13 Jan 2023, 08:59 PM

Originally posted by Frank Rogers View Post

I'm playing around with a merge record sort routine. I have one routine to sort by PartNo, one to sort by Description, one to sort by QuantityOnHand, ... It works with a hard coded field, but if I want to sort by another field it requires another sort routine which is 99% the same. I'm trying to knock down many routines into one general routine.

I could set up an equate to each field and then a CASE SELECT to pick the field but that feels messy.

I need to pass the starting field location and field length to the routine.

With the UDT I can just @pString[RecNo] = @pString[RecNo +1] but the compare is what is hanging me up

,l'm pretty sure that the time difference between your Merge Sort and an ArraySort on your UDTs would be negligible in practice
As for needing different routines, this is fairly simple and VERY easy to understand and maintain

'

Code:

#COMPILE EXE
#DIM ALL

TYPE MyRecord
  PartNo AS STRING * 20
  Desc AS STRING * 50
  Quantity AS LONG
END TYPE
GLOBAL gsInvSortOrder AS STRING

%NoOfItems = 10000

FUNCTION PBMAIN() AS LONG
    LOCAL lDebug AS LONG: TXT.WINDOW EXE.FULL$, 200,50,40,100 TO lDebug
    LOCAL i AS LONG
    LOCAL t AS DOUBLE
    DIM arrInv(1 TO %NoOfItems) AS MyRecord
    FOR i = 1 TO UBOUND(arrInv)
        arrInv(i).PartNo = "Part No" & STR$(i)
        arrInv(i).Desc = MID$("abcdefghijklmnopqrstuvwxyz",(i MOD 26) +1,1) & "-Description of Part No " & STR$(i)
        arrInv(i).Quantity = RND(0 ,1000)
    NEXT
    TXT.PRINT "Array of " & STR$(%NoOfItems) & " inventory itmes created"
    DO
        TXT.PRINT
        TXT.PRINT "Sort by [P]art Number,[D]escription,[Q]uantity (any other key to exit)
        TXT.WAITKEY$ TO gsInvSortOrder
        IF INSTR("PDQ",UCASE$(gsInvSortOrder)) = 0 THEN EXIT DO
        t = TIMER
        ARRAY SORT arrInv(),CALL SortArray
        t = TIMER - t
        TXT.PRINT
        TXT.PRINT "Sorted " & STR$(%NoOfItems)  & " items in " & FORMAT$(t,"0.000") & " seconds"  & " By " & UCASE$(gsInvSortOrder)
        SHowData(arrInv())
    LOOP

'Finalise
    TXT.COLOR = %RGB_BLUE
    TXT.PRINT
    TXT.PRINT "  ....Press any key to exit": TXT.WAITKEY$: TXT.END
END FUNCTION

FUNCTION SHowData(a() AS MyRecord) AS LONG
    LOCAL i AS LONG
    FOR i = 1 TO 3
        TXT.PRINT a(i).partno, a(i).desc,a(i).quantity
    NEXT
    TXT.PRINT
    FOR i = UBOUND(a()) - 3  TO UBOUND(a())
        TXT.PRINT a(i).partno, a(i).desc,a(i).quantity
    NEXT
 END FUNCTION

FUNCTION SortArray(p1 AS MyRecord,p2 AS myRecord) AS LONG
     SELECT CASE UCASE$(gsInvSortOrder)
         CASE "P"
             IF p1.PartNo < p2.PartNo THEN FUNCTION = -1 :EXIT FUNCTION 
             IF p1.PartNo > p2.PartNo THEN FUNCTION =  1
         CASE "D"
             IF p1.Desc < p2.Desc THEN FUNCTION = -1 :EXIT FUNCTION 
             IF p1.Desc > p2.Desc THEN FUNCTION =  1

         CASE "Q"
             IF p1.Quantity < p2.Quantity THEN FUNCTION = -1 :EXIT FUNCTION 
             IF p1.Quantity > p2.Quantity THEN FUNCTION =  1
     END SELECT

END FUNCTION
'

**Michael Mattias** · 15 Jan 2023, 01:00 PM

MIght this help you?

Code:

UNION  MultiPointer
     A  AS STRING PTR * 2
     B  AS STRING PTR * 3
     C  AS STRING PTR * 4
     ....
END UNION

It's a thought.

'ARRAY SORT FROM tartpos TO endpos' sounds like it might do the job for you. Just pass in the startpos and endpos to a procedure.

Or pass to a MACRO kind of like I demo'd here..

CC3+/Win7+: ARRAY SORT UDT array on member name September 04, 2002

That's another thought.

**Frank Rogers** · 15 Jan 2023, 06:09 PM

Good suggestion Michael. I played with FROM and TO in ARRAY SORT but it didn't quit flow as smoothly as I liked so I went off in search of making my own. My own wasn't as flexible, so I scraped it for ARRAY SORT Call... It's a tiny bit slower than a hand built sort but flexibility wins this fight.

**Dale Yarker** · 16 Jan 2023, 02:19 AM

Code:

UNION MultiPointer
A AS STRING PTR * 2
B AS STRING PTR * 3
C AS STRING PTR * 4
....
END UNION

Its not a thought, its nonsense! You still must declare the string as fixed type with a length.

Declare string as longest possible, and get VARPTR of the string. Then use PEEK$() and POKE$ instead of @. You still have to be careful about not accidentally over-writing other variables, and increment the pointer by declared length.

Cheers,

**Michael Mattias** · 16 Jan 2023, 02:33 PM

Its not a thought, its nonsense! You still must declare the string as fixed type with a length.

Depending on the application, it might be exactly what he needs. (How the data are actually stored is not shown)

In a more general sense, I've said for a long time if you don't have some bad ideas, you don't have enough Ideas.

But it really looks like...

Code:

FUNCTION |MACRO SortMyUDTArray (array(), Offset , length)

.. is going to be the way to go.

**Frank Rogers** · 16 Jan 2023, 04:45 PM

Originally posted by Frank Rogers View Post

... It's a tiny bit slower than a hand built sort but flexibility wins this fight.

Announcement

How to set the Pointer size at runtime

How to set the Pointer size at runtime

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment