Re: [Boost-Users] Query regarding the use/limitiations of traits in metaprogramming

It looks like it's working find with gcc-2.95.3-5. One thing, though, it gave me a compiler error when I didn't #include <typeinfo>. Since you didn't list which header files you're including, I'm not sure if you're including <typeinfo> or not and what the Borland's behaviour would be if you didn't. Noel --- hicks wrote:

Hello

...

I have some questions for you: 1. Why are you so interested in the value of "typeid.name()"?

I'm not particulary interested in typeid::name() as an end result. As I said in the letter, I have real task at hand, and condensed the code so that anybody looking at it would be able to get to the point quickly. I used typeid()::name() here as a means of demonstating the problem. The result which is shown by typeid()::name() corresponds to the result encountered in the real task, namely, the next stage of processing is supposed to take place in traits1 but takes place in traits1 instead. In other words, the branching mechanism fails silently.

...

2. Are you sure this isn't a compiler problem? What compiler(s) have you tried this on?

Borland Builder C++ 5. Unfortunately, it's the only one to which I have access now. It could well be a compiler problem. Unless there are some rules which I don't understand. If it is a compiler problem perhaps someone knows of a workaround. I have tried various approaches to no avail.

Craig Hicks

Noel Yap wrote:

...

--- hicks mailto:hicks@kgk.co.jp wrote:

...

Query regarding the use/limitiations of traits in metaprogramming

I am trying to use traits pattern to assist in metaprogramming. I have the following problem.

A traits class "trait" defers a decision to another traits class "trait1" depending on the result of a third traits class "traita". However, the compiler does not cooperate. The code below excutes but gives unexpected output (also shown).

[Note:: The code has been reduced to the minimal to demonstrate the problem. I have a real task at hand not shown here.]

template <class T> struct traita { BOOST_STATIC_CONSTANT(int, value=0); }; template <> struct traita<int> { BOOST_STATIC_CONSTANT(int, value=1); };

template struct trait1 {};

template <class T> struct trait { BOOST_STATIC_CONSTANT(int, value=traita<T>::value); typedef trait1< ;T, value> t1; };

int main(int argc, char* argv[]) { std::cout << typeid(trait<int>::t1).name() << std::endl << trait<int>::value << std::endl ; return 0; }

output::

trait1 1

It would be nice if the first line of output were "trait1", but it isn't.

Any assistance would be greatly appreciated.

I have some questions for you: 1. Why are you so interested in the value of "typeid.name()"? 2. Are you sure this isn't a compiler problem? What compiler(s) have you tried this on?

Noel

__________________________________________________ Do You Yahoo!? Yahoo! Health - your guide to health and wellness http://health.yahoo.com

Info: http://www.boost.org Wiki:

http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl

...

Unsubscribe: mailto:boost-users-unsubscribe@yahoogroups.com

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

[Non-text portions of this message have been removed]

------------------------ Yahoo! Groups Sponsor

Info: http://www.boost.org Wiki:

http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl

Unsubscribe: mailto:boost-users-unsubscribe@yahoogroups.com

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

__________________________________________________ Do You Yahoo!? Yahoo! Health - your guide to health and wellness http://health.yahoo.com