allow lexical_cast to handle based number strings

[Steven Robbins <smr@…>]

This report comes from Debian user "Wesley J. Landaker" <wjl@…> See ​http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=412931

Please allow lexical_cast to handle based number strings. Currently, it only supports decimal strings, but it should be able to handle octal and hexadecimal strings if they have correct prefix notations.

Background: This happens because lexical_cast constructs a stringstream, which by default has the ios_base::dec flag set. If this flag is unset, it will still read decimal strings, but will also automatically determine the base by looking at the prefix, so will read decimal strings if the number starts with 0-9, octal if it starts with 0, and hex if it starts with 0x.

Fix: After constructing the stringstream, simply call stringstring::unsetf(std::ios_base::basefield). Because decimal numbers are still handled correctly, this is totally backwards compatible.

[nasonov]

This feature is out scope of the library.

References:

[1] Boost documentation

http://www.boost.org/libs/conversion/lexical_cast.htm#motivation

[2] Lexical Conversion Library Proposal for TR2

​http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2006/n1973.html

Both have this text: For more involved conversions, such as where precision or formatting need tighter control than is offered by the default behavior of lexical_cast, the conventional stringstream approach is recommended.

[René Rivera]

I've run into this problem many, many times. And hadn't realized it was possible to provide for better number conversion. In my opinion allowing such number base conversion is not only within the scope, but by not allowing the conversions it specifically imposes an artificial "tighter control" than the expected behavior.

Any chance we can change your mind about this?