ARRAY-ideas

not a completely new idea (mentioned already in some other thread) but i would like to use the powers of Array Scan on data stored at heap-memory.

How? Nearly the same way it works for arrays of dynamic strings currently. As far as i know a dynamic-string-array is just an array of stringpointers - right?

I want to have an array of heap-pointers and scan the content of the pointed heap-areas.
small example:

Code:

---

DWord myPtrs = HEAP_AllocByStr(MKDWD$( _
          HEAP_AllocByStr("this"), _
          HEAP_AllocByStr("is an"), _
          HEAP_AllocByStr("array")  )      )
 
Dim vPtr(HEAP_Size(myPtrs)/4) As DWord At myPtrs

' tell somehow its an array of pointers
' so i simply insert keyword Ptr here:
 
PrintL Array Ptr Scan vPtr, Collate Ucase, = "ARRAY"

' should print 3 now

---

PS. if a pointer contains 0 it should not throw error-
and btw. change the behaviour of heap-memory and disable allocating sizes of less than 1 byte/ empty strings!

________________________________________________________________________

Oh yes- another Array-idea were to have some way to loop through all elements of an array without the need of additional counter-variables.
something alike

Code:

---

With [Each Element Of] Array myArray
'...do something...
' would be limited to arrays of udt only
End With


' or another way:
For Array myArray
  ' now myArray is an overlay at current element!
  ' (no limitation as above, so can use arrays of byte, long etc)
Next

---

(see more detailed examples for the second way here)

Rene,

I can only imagine what you could do with thinBASIC if you had access to the source. Maybe it's time Eros opens the door to serious advocates of the language. If that isn't a choice Eros is willing to make, you should find yourself an open source project and make these contributions count.

John

@John,
it is admirable how you always try to bring people on your side every time someone suggest new features not directly present into a programming language.
I hope you are payed for this job.
Well done :eusaclap:


@Rene,
I will check if I can add the first request. It is very very specific command. I would like something more general to be used by the average programmer. Will see.
Regarding the second request, I already started a FOREACH command but not yet finished.

Ciao
Eros

thanks for your reply... makes me feel tB is collecting energy for an upcoming boost in future. Actually the first (Array Scan)-request is the more important one to me- there a quite a few ways to loop through an array available already - the suggest was more because it's possible and fits the topic.

The other thing - inbetween the lines - seems the most important to me - make Heap_Size < 1 unavailable please. I always use Heap_Size to detect if Heap at a certain pointer exists at all since there's no Heap_Exists and if a Heap_Size can not be below 1 then it would not be necessary to check any else than size to test the existence/validity of some heap-memory.
to delete some old memory i check mostly like that since there's no other method:

Code:

---

if Heap_Size(x) then
 Heap_Free(x)
 x = 0
endif

---

but if Heap_Size(x) = 0 then i have a lost orphan pointer and some 0-len-memory-block stays allocated. So to me seems very important that Heap_Size of 0 Bytes are not accepted/allocated and simply return 0 on Heap_Alloc/Heap_ReAlloc/Heap_AllocByStr if size of data < 1

Quote:

---

@John,
it is admirable how you always try to bring people on your side every time someone suggest new features not directly present into a programming language.
I hope you are payed for this job.
Well done :eusaclap:

---

Excuse me but where did you read me asking Rene to join me or the Script BASIC project? I said find an open source project. Please don't start reading between the lines. There is already enough of that going around.

A simple answer how you feel about opening the source would have been sufficient.

Ciao Renè,

as a starter, please find here enclosed an updated thinCore.dll.
To install just unzip it into \thinBasic\ directory replacing your current thinCore.dll.

New syntax for HEAP_AllocByStr is the following:
ptr = HEAP_AllocByStr(String [, AllowEmptyStringAllocation])

In case of empty string, now allocation will take place (function will return zero) unless optional param "AllowEmptyStringAllocation" will be %TRUE


Also implemented Heap_End as requested at http://www.thinbasic.com/community/p...hp?issueid=471


Actually I'm finishing to develop and document two new thinBasic modules: Excel and AdoDB. Quite excited about that.
As soon as I will finish (documentation included) I will publish thinBasic 1.9.14. Around end of next week.

In the meantime I will think about your recent requests. "Array Ptr Scan ..." is quite easy, maybe I will be able to include in this round.

Ciao
Eros

Ciao again Renè.

Attached a new thinCore.dll where I've developed a preliminar version of your ARRAY SCAN ... PTR ... request.

Syntax is the following (PTR just after array name):

Code:

---

Uses "console"

'---Allocate a string that is a sequance of some DWORDs.
'---Each DWORD represents a PTR to a heap memory area where a string is allocated
DWord myPtrs = HEAP_AllocByStr( _
                                MKDWD$( _
                                        HEAP_AllocByStr("this")  , _
                                        HEAP_AllocByStr("is an")  , _
                                        HEAP_AllocByStr("array")  , _
                                        HEAP_AllocByStr("of")    , _
                                        HEAP_AllocByStr("some")  , _
                                        HEAP_AllocByStr("strings")  _
                                      ) _
                              )


'---Define a virtual array over-imposing its structure to the previous allocated pointer
Dim vPtr(HEAP_Size(myPtrs)/SizeOf(DWord)) As DWord At myPtrs
 
'---Now scan that area using the new PTR option.
'---PTR tells to ARRAT SCAN that the DWORD is in reality a pointer to some memory area
'---where to find a heap allocation of a string to be used for comparison. 
'---ATTENTION: this will work only for DWORD arrays!!!!!!!!!!!!!!!!!!!!!!!
PrintL Array Scan vPtr Ptr, Collate Ucase, = "SOME"


PrintL "Press any key"
WaitKey

---

Let me know if you find any bug. I didn't make much tests.

Ciao
Eros

cooool - very excited - it feels like being a child having birthday :D
- still testing and trying -

Heap_End also very nice working - now can virtually go "For Each" without the need of allocating additional variables nor calculations

Code:

---

Uses "console"
 
DWord myPtrs = HEAP_AllocByStr( _
                                MKDWD$( _
                                        HEAP_AllocByStr("this")  , _
                                        HEAP_AllocByStr("is an")  , _
                                        HEAP_AllocByStr("array")  , _
                                        HEAP_AllocByStr("", TRUE) , _
                                        HEAP_AllocByStr("")      , _
                                        HEAP_AllocByStr("some")  , _
                                        HEAP_AllocByStr("strings")  _
                                      ) _
                              )
 
'test Heap_End
Dim vSurf As DWord At myPtrs

While VarPtr(vSurf) < Heap_End(myPtrs)
  PrintL Hex$(vSurf,8) & " : " & HEAP_Get(vSurf)
  SetAt( vSurf, VarPtr(vSurf) + SizeOf(vSurf))
Wend

PrintL $CRLF & Repeat$(42, "*")
PrintL $CRLF & "Any key to end"
WaitKey

---

Got an Error! But not related to the current topic but while experimenting with new tB-functions.
This is my testing script, requires the heap-unit below and thinCore.dll above anyway since the unit makes use of the new Array Scan Ptr-functionality

Code:

---

Uses "console"
#INCLUDE "Heap.tBasicU"
     
' some simple udt for test
Type t_vec3d
  X As Double
  Y As Double
  Z As Double
  SetPos As Function
End Type           

Function t_Vec3d.SetPos(ByVal X As Double, ByVal Y As Double, ByVal Z As Double)
  Me.X = X
  Me.Y = Y
  Me.Z = Z
End Function
 
' ---------------
 
Dim foo As Heap

' create some space to store different things at foo
foo.AllocLike("Heap", 3)

' first heap at foo shall contain a string
Dim vHeap Like foo.GetType At foo.pData
vHeap.Be("Hello World", "String")

' next heap at foo shall contain some longs
SetAt(vHeap, VarPtr(vHeap) + SizeOf(vHeap))
vHeap.Be(MKL$(1,2,3,4,5), "Long")

' next heap at foo shall contain 3 vecs
SetAt(vHeap, VarPtr(vHeap) + SizeOf(vHeap))
vHeap.allocLike("t_Vec3d", 3)
  ' fill in some udt-data
  Dim vVec Like vHeap.GetType At vHeap.pData
  vVec.SetPos(1.1, 1.2, 1.3)
  SetAt( vVec, VarPtr(vVec)+ SizeOf(vVec) )
  vVec.SetPos(2.1, 2.2, 2.3)
  SetAt( vVec, VarPtr(vVec)+ SizeOf(vVec) )
  vVec.SetPos(3.1, 3.2, 3.3)
 
' now reset to first heap
SetAt( vHeap, foo.pData )
While VarPtr(vHeap) < foo.GetEnd

  PrintL "Type is " & vHeap.GetType
 
  ' replace below
  '  "HEAP_Get(vHeap.pType)" with "vHeap.GetType()" to have error
  'the following line is the one:
  Select Case HEAP_Get(vHeap.pType)
    Case "STRING"
      PrintL "contains :" & vHeap.Is 
    Case "LONG"
      Dim vLong As Long At vHeap.pData
      While VarPtr(vLong) < vHeap.GetEnd
        PrintL vLong
        SetAt(vLong, VarPtr(vLong) + SizeOf(Long))
      Wend
    Case "T_VEC3D"
      SetAt(vVec, vHeap.pData)
      While VarPtr(vVec) < vHeap.GetEnd
        PrintL "X:" & vVec.X
        PrintL "Y:" & vVec.Y
        PrintL "Z:" & vVec.Z
       
        SetAt(vVec, VarPtr(vVec) + SizeOf(vVec))
      Wend
  End Select
  ' push forward to next heap 
  SetAt(vHeap, VarPtr(vHeap) + SizeOf(vHeap) )
Wend

PrintL $CRLF & Repeat$(42, "*")
PrintL $CRLF & "Any key to end"
WaitKey

---

check line 53 of the script.

Code:

---

Select Case [some type-function-result]

---

results in crash, no matter if parenthesis used or not. I don't know if that is always the case since i discovered this as a reason of error for the first time

Rene,

I think the crash is related to the fact I never adjusted
SELECT CASE <...>
when <...> is a UDT function.

When SELECT CASE <...> is fired, ThinBASIC try to understand if the next expression (the one inside <...>) is a numeric expression or a string expression.
After that it is able to know that all internal CASE ... are one type or the other.

My idea is that direct UDT functions in a SELECT statements are not supported and for this reason it does not know that vHeap.GetType will return a string.

It should be an easy FIX but ... :(
Actually I'm still at work after 14 hours working.
I still have 1 hour drive at home.
If I will have the forces I will check this night or (maybe) tomorrow, sorry.

Ciao
Eros