[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #6 from Paolo Carlini paolo.carlini at oracle dot com --- Thanks Janis. In C++11 we have *explicit* conversion operators. Would they help? A safe approach woould providing the operators only in c++11 mode.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815
--- Comment #7 from Paolo Carlini paolo.carlini at oracle dot com ---
Something like:
Index: include/decimal/decimal
===
--- include/decimal/decimal(revision 203915)
+++ include/decimal/decimal(working copy)
@@ -250,8 +250,10 @@
/// Conforming extension: Conversion from scalar decimal type.
decimal32(__decfloat32 __z): __val(__z) {}
-// 3.2.2.5 Conversion to integral type. (DISABLED)
-//operator long long() const { return (long long)__val; }
+#if __cplusplus = 201103L
+// 3.2.2.5 Conversion to integral type.
+explicit operator long long() const { return (long long)__val; }
+#endif
// 3.2.2.6 Increment and decrement operators.
decimal32 operator++()
@@ -333,8 +335,10 @@
/// Conforming extension: Conversion from scalar decimal type.
decimal64(__decfloat64 __z): __val(__z) {}
-// 3.2.3.5 Conversion to integral type. (DISABLED)
-//operator long long() const { return (long long)__val; }
+#if __cplusplus = 201103L
+// 3.2.3.5 Conversion to integral type.
+explicit operator long long() const { return (long long)__val; }
+#endif
// 3.2.3.6 Increment and decrement operators.
decimal64 operator++()
@@ -417,8 +421,10 @@
/// Conforming extension: Conversion from scalar decimal type.
decimal128(__decfloat128 __z): __val(__z) {}
-// 3.2.4.5 Conversion to integral type. (DISABLED)
-//operator long long() const { return (long long)__val; }
+#if __cplusplus = 201103L
+// 3.2.4.5 Conversion to integral type.
+explicit operator long long() const { return (long long)__val; }
+#endif
// 3.2.4.6 Increment and decrement operators.
decimal128 operator++()
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #8 from Daniel Krügler daniel.kruegler at googlemail dot com --- (In reply to Paolo Carlini from comment #6) I also would like to encourage using explicit conversion functions. This is explicitly suggested in the updated C++11 integration document: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2012/n3407.html
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #9 from Janis Johnson janis at gcc dot gnu.org --- I haven't paid attention to decimal float since leaving IBM, so it was very interesting to see the updated C++11 working paper. It makes sense to me to use C++11 functionality in the GNU C++ decimal float support, especially in the case of explicit conversions.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 Paolo Carlini paolo.carlini at oracle dot com changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2013-10-22 Assignee|unassigned at gcc dot gnu.org |paolo.carlini at oracle dot com Ever confirmed|0 |1 --- Comment #10 from Paolo Carlini paolo.carlini at oracle dot com --- Ah good, thanks Janis. Thus let's add in C++11 mode these explicit operators with a comment pointing to n3407. More than that is beyond the scope of this PR but it would be nice if somebody volunteers to follow the topic, prepares patches, etc.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815
--- Comment #11 from ulf q1 at oxyba dot de ---
i see evil implicit conversion.
i modified lines 254, 337 and 421 to
explicit operator long long() const { return (long long)__val; }
like suggested in n3407.
#includedecimal/decimal
int main(){
std::decimal::decimal32 x(101.5051);
long long a = (long long)x; // OK
long long b = static_castlong long(x); // OK
//long long c = x; // error
//long d = (long)x; // error
//long e = static_castlong(x); // error
std::decimal::decimal64 y(101.5051);
long long f = (long long)y; // OK
long long g = static_castlong long(y); // OK
//long long h = y; // error
//long i = (long)y; // error
//long j = static_castlong(y); // error
std::decimal::decimal128 z(101.5051);
long long k = (long long)z; // OK
long long l = static_castlong long(z); // OK
//long long m = z; // error
//long n = (long)z; // error
//long o = static_castlong(z); // error
return 0;
}
some other issues regarding std::decimal
std::is_decimal_floating_pointstd::decimal::decimalXXX in type_traits
see page 7, chapter 3.11.1 and 2 (page 62) in
www.open-std.org/jtc1/sc22/open/n4112.pdf
std::numeric_limitsstd::decimal::decimalXXX in limits
see chapter 3.3 in
http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2006/n1977.html
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #12 from Janis Johnson janis at gcc dot gnu.org --- I obviously don't know C++ very well and the decimal float support in libstdc++ is very ugly. It would be nice if someone rewrites it in actual C++ someday; the tests should help with that effort.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #13 from Paolo Carlini paolo.carlini at oracle dot com --- Thus, let's make a decision: either this is a bug, importance minor as submitted, about conversions to integer, as appeared to be, or it's a catch all bug about all the issues with the TR decimal + all the work toward a C++17 standard facility.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #2 from Paolo Carlini paolo.carlini at oracle dot com --- In any case, keep in mind ulf that we are looking for help in this area: for various reasons, the code isn't really maintained these days (eg, the original implementor moved to another job, etc) Thus, contributions welcome!
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815
ulf q1 at oxyba dot de changed:
What|Removed |Added
CC||janis187 at us dot ibm.com
--- Comment #3 from ulf q1 at oxyba dot de ---
i checked the code in the file decimal (/usr/include/c++/4.7/decimal/). The
programmer already considered conversion to integers. however the lines are
commented.
for std::decimal::decimal32
...
line 253 // 3.2.2.5 Conversion to integral type. (DISABLED)
line 254 //operator long long() const { return (long long)__val; }
...
for std::decimal::decimal64
...
line 336 // 3.2.3.5 Conversion to integral type. (DISABLED)
line 337 //operator long long() const { return (long long)__val; }
...
for std::decimal::decimal128
...
line 420 // 3.2.4.5 Conversion to integral type. (DISABLED)
line 421 //operator long long() const { return (long long)__val; }
...
i uncommented it and my problems were gone. but for some reason the code is
disabled on purpose. what is the reason for it?
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #4 from Janis Johnson janis at gcc dot gnu.org --- I vaguely remember a problem with those conversion and have no idea now what it was, but I'll try to find something in the mailing list archives. decimal/decimal includes a workaround: /// Non-conforming extension: Conversion to integral type. long long decimal32_to_long_long(decimal32 __d); long long decimal64_to_long_long(decimal64 __d); long long decimal128_to_long_long(decimal128 __d); long long decimal_to_long_long(decimal32 __d); long long decimal_to_long_long(decimal64 __d); long long decimal_to_long_long(decimal128 __d);
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 --- Comment #5 from Janis Johnson janis at gcc dot gnu.org --- From Janis in http://gcc.gnu.org/ml/libstdc++/2009-09/msg00115.html The TR defines, within each of the classes decimal32, decimal64, and decimal128, the following conversion: // 3.2.[234].4 conversion to integral type: operator long long() const; Apparently there's a way to implement this if one knows enough about C++, but providing this implicit conversion from a decimal float type to long long allows further implicit conversions from long long to other types, including float, double, and long double, which are not supposed to be allowed. Furthermore, a conversion from a decimal float type to a generic float type that goes through long long truncates the fractional part of the value which is rather surprising. This version of the patch leaves out that conversion and provides a set of functions to convert to long long, which made testing the rest of the functionality much easier. Any suggestions for how to do with using the current (not C++0x) standard? Or is it OK to require the use of C++0x functionality with this extension? From Benjamin Kosnik in http://gcc.gnu.org/ml/libstdc++/2009-09/msg00127.html I think it is not ok to require the use of C++0x with the decimal floating point extension. It is clearly designed with C++2003 in mind, and makes no use of C++0x features. When you post testsuite files demonstrating this issue in detail I'll take a look at this overload issue. From Janis in http://gcc.gnu.org/ml/libstdc++/2009-10/msg4.html With the submission on Monday I mentioned that this patch does not support an implicit conversion to integral types (long long) from each of the decimal classes. Those are in the patch, commented out and with a comment that now says (DISABLED). If those are not disabled then all of the error checks in the bad-*.cc tests fail. I can't see how to allow implicit conversions to long long while also prohibiting bitwise operations with decimal floating-point operands or conversions and casts to generic floating-point types. I didn't find any further discussion of this issue.
[Bug c++/58815] Casting/Conversion operator for std::decimal not supported
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58815 Daniel Krügler daniel.kruegler at googlemail dot com changed: What|Removed |Added CC||daniel.kruegler@googlemail. ||com --- Comment #1 from Daniel Krügler daniel.kruegler at googlemail dot com --- The notion of throws an error is somewhat misleading in this context. What you intend to describe is that the compiler rejects the code.
