• =?ISO-8859-1?Q?Julio_C=E9sar_Carrascal_Urquijo?= Mar 04 2011
• spir <denis.spir gmail.com> Mar 04 2011
• Walter Bright <newshound2 digitalmars.com> Mar 04 2011

=?ISO-8859-1?Q?Julio_C=E9sar_Carrascal_Urquijo?= writes:

I was writing something like the following code:

// double y1, y2, width;
size_t index = to!size_t((x2 - x1) / width);
if (index >= 0 && index < grid.length) {
     ...
} else {
     ...
}

When I finished writing the condition I remembered that size_t is 
actually an uint so it can't be negative.

Just for fun I tested this:

writeln(to!size_t(-1));

std.conv.ConvOverflowException std\conv.d(8): Error: overflow Conversion 
negative overflow

That's not what I was expecting but it's OK. For my use case I changed 
the code use int but, with 64bit already here, that's not quite correct.

So if I may suggest: Could we have a signed size_t in Phobos? ssize_t or 
with another name.

I don't want to replace size_t. I just want a signed type that's correct 
for storing an index for an array in 32 and 64 bit.

PS: I realize this has been discussed before. I just hope we get some 
luck this time. <g>

Thanks

spir <denis.spir gmail.com> writes:

On 03/04/2011 09:54 PM, Julio César Carrascal Urquijo wrote:
 I was writing something like the following code:

 // double y1, y2, width;
 size_t index = to!size_t((x2 - x1) / width);
 if (index >= 0 && index < grid.length) {
 ...
 } else {
 ...
 }

 When I finished writing the condition I remembered that size_t is actually an
 uint so it can't be negative.

 Just for fun I tested this:

 writeln(to!size_t(-1));

 std.conv.ConvOverflowException std\conv.d(8): Error: overflow Conversion
 negative overflow

 That's not what I was expecting but it's OK. For my use case I changed the code
 use int but, with 64bit already here, that's not quite correct.

 So if I may suggest: Could we have a signed size_t in Phobos? ssize_t or with
 another name.

 I don't want to replace size_t. I just want a signed type that's correct for
 storing an index for an array in 32 and 64 bit.

 PS: I realize this has been discussed before. I just hope we get some luck this
 time. <g>


unittest {
     size_t i;
     sizediff_t n;
     writefln("%s %s", typeid(i), typeid(n));
     // --> "uint int" on 32-bits
}

Denis
-- 
_________________
vita es estrany
spir.wikidot.com

Walter Bright <newshound2 digitalmars.com> writes:

Julio César Carrascal Urquijo wrote:
 Could we have a signed size_t in Phobos? ssize_t or 
 with another name.


ptrdiff_t