TStringList.SetCapacity related
lazarus / 64-FPC development

----------------------------------------------------------------------
lagprogramming · 2015-10-27 19:19

TStringList.SetCapacity can be found inside stringl.inc.

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i Procedure
TStringList.SetCapacity(NewCapacity: Integer);Var NewList : Pointer; 
 MSize : Longint;begin  If (NewCapacity<0) then     Error
(SListCapacityError,NewCapacity);  If NewCapacity>FCapacity then   
begin    GetMem (NewList,NewCapacity*SizeOf(TStringItem));    If
NewList=Nil then      Error (SListCapacityError,NewCapacity);    If
Assigned(FList) then      begin     
MSize:=FCapacity*Sizeof(TStringItem);      System.Move
(FList^,NewList^,MSize);      FillWord
(Pchar(NewList)[MSize],(NewCapacity-FCapacity)*WordRatio,
0);//<--------------      FreeMem (Flist,MSize);      end;... 
   I'd like the rationale to be checked. I think somebody tried to
optimize the line
Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i
FillByte(Pchar(NewList)[MSize],(NewCapacity-
FCapacity)*SizeOf(TStringItem),0);with
Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i FillWord
(Pchar(NewList)[MSize],(NewCapacity-FCapacity)*WordRatio, 0);
   The above lines don't fill with zeros the same size, the existing
line fills only half the space difference. This is because
classesh.inc has {$H+}, which makes
SizeOf(TStringItem)=2*SizeOf(Pointer), while WordRatio =
SizeOf(Pointer) Div SizeOf(Word).
   I don't think it's very important but, am I wrong?

----------------------------------------------------------------------
engkin · 2015-10-28 04:03

Quote from: lagprogramming on October 27, 2015, 08:19:25 pm   I'd like
the rationale to be checked.

[...]

   The above lines don't fill with zeros the same size, the existing
line fills only half the space difference.

[...] am I wrong?

No, you are right!

Edit:
Maybe FillDWord was meant to be used instead of FillWord?

----------------------------------------------------------------------
Bart · 2015-10-28 11:20

File a bugreport, otherwise it probably won't get noticed.

Bart

----------------------------------------------------------------------
lagprogramming · 2015-10-28 15:40

I know 16-bit is a valid target. For knowledge, will FillDWord work
fine for users that want to target msdos i8086?
I agree with Bart, somebody might fill a Mantis bug entry.

----------------------------------------------------------------------
engkin · 2015-10-28 17:05

Quote from: lagprogramming on October 28, 2015, 04:40:42 pmwill
FillDWord work fine for users that want to target msdos i8086?

It is in generic.inc file, which according to system.inc comment:
Quote{ Include generic pascal only routines which are not defined in
the processor
  specific include file }
{$I generic.inc}        

So it should work ok.

----------------------------------------------------------------------
christian1987 · 2015-10-29 08:37

I was trying to get this to crash... couldn't.

Here is why:
1132      Pointer(Flist^[Index].Fstring):=Nil;  // Needed to
initialize.

TStringList.InsertItem sets it to nil AGAIN.

So I think maybe FillWord code is unneccessary? Not sure.