Hello

I think i've found a problem with respect to big integers (int64,qword types). Please look at the following testprogram and tell me if there's a patch. I'm using fpc version 1.0.4 on Win32.

program Bugtest(input,output);
var
   big : int64;
function intpower(b,p : longint) : int64;
var i : longint;
begin
   intpower:=1;
   for i:=1 to p do intpower:=b*intpower;
end;
begin { main }
   big := 4294967296;   { =2^32 }
   writeln(' 2^32=',big);
   writeln(' 2^32=',intpower(2,32));
end.

This program gives the following output:

2^32=0
2^32=4294967296

Apparently something has gone wrong putting
the value 2^32 into 'big'

Please help me!

Greetings Jan Feitsma

Try a 1.1 development snapshot. The 1.1 development series supports constant values larger than maxlongint better

Jan Feitsma wrote

program Bugtest(input,output); var    big : int64;

[snip]

  begin { main }    big := 4294967296;   { =2^32 }

The problem is that 1.0.x only supports 32bit constants. The current 1.0.5 gives an error on this line:

bugtest.pp(13,11) Error: Incompatible types: got "S80REAL" expected
"INT64"

In 1.1, support for 64 bit constants has been added and there your program works correctly.

Jonas