Tcl Source Code

Found problem with "expr" while doing calculations 
For example : 
% expr (576.64 - 554.176) * 1.0 / 0.576
39.0
% expr int((576.64 - 554.176) * 1.0 / 0.576)
38 

How come the int for 39.0 will become 38?

The issue is that the calculation isn't yielding the value that you think it is; when I try with Tcl 8.5 or 8.6 (both of which are much more careful in the rendering of floating point numbers than 8.4 and before were), I get this:

% expr (576.64 - 554.176) * 1.0 / 0.576
38.9999999999999

The one interesting thing about Tcl is that we now guarantee to use the minimum number of digits to represent a number when converting from the floating point internal representation to the decimal string form, subject to the constraint that the reverse operation (string-to-IEEE-double) gives exactly the same thing. That is a non-trivial guarantee; it takes quite a lot of code in Tcl's implementation to achieve it. If you really want to use a smaller number of digits than what is properly necessary, the format command is the right approach, though you should be aware that this will do some rounding.