Re: [Bug-apl] clang Warning - revisited

[Juergen Sauermann]

Hi Peter,

I have changed it in SVN 215.

I still disagree with the warning because in a comparison of an enumeration member
with an integer should perform integer promotion of the enum member to int and then do
integer comparison and NOT convert the int to an enum member (which it can't) and then
do the comparison (or issue the warning).

In other words, I read "l < 10" as "(int)l < 10" (which can be true or false) and not as "l < (Function_line)10"
which can also be true or false because if I allow implicit (Function_line)10 then I cannot rule out (Function_line)42
either.

The really interesting question is if the compiler would optimize "if (i < 10)" away which I would consider a fault
in the compiler. As a warning it is just a bit annoying because it forces "default: ;" all over the place.

/// Jürgen



On 04/19/2014 02:36 AM, Peter Teeson wrote:

Hi Jürgen:

I've been thinking again about this warning and I'm inclined to agree with it.

..MyProjects/GNUAPL/apl-svn/src/UserFunction.cc:1191:10: Comparison of constant 10 with _expression_ of type 'Function_Line' is always true

   if (l < 10)   ucs.append(UNI_ASCII_SPACE);

These suggestions eliminate the Warning.

In File APL_Types.hh

enum Function_Line

{

   Function_Retry  = -2,   // →'' in immediate execution

   Function_Line_0 = 0,

   Function_Line_1 = 1,

   Function_Line_MAX = 10  // << === My suggestion

};

In UserFunction.cc:1191:10:

   if (l < Function_Line_MAX)   ucs.append(UNI_ASCII_SPACE); // << === My suggestion