can anyone tell me why this will work? unsigned char* a; double b= 45456464645.43564564564; a= reinterpret_cast(&b); double c= reinterpret_cast(*a); so c at the end equal the exact value of b so this IS okay.. but why this time this can''t be done? unsigned char* a; char b[50]= "0123456789012345678901234567890123456789012345678"; a= reinterpret_cast(&b); std::string c= reinterpret_cast(*a); it do a 1st chance exception when a put value of a to c..? by the way i''m using Visual Studio 6... i''ve tried the same code in Borland C++ 5 and... well it seem to work can anyone explain me the problem??!?

That only works when b and c are the same type.

A reinterpret_cast will do really odd things when trying to treat a character array as a std::string. Do you even know what reinterpret_cast does??

Don't listen to me. I've had too much coffee.

[edited by - sneftel on October 22, 2002 3:43:18 PM]

well i''wasn''t knowing there was danger doing this with char array & string... and how would you do this then? string or array?

and by the way Sneftel i know that reinterpret_cast is used to convert a pointer to another one by doing NO test if it''s possible, it simply do it, so doing this, you can cause a lot of problem if not used correctly..., this is what i know about it, fell free to correct me or add any comments on what it do

Right. And treating a character array as a string is one of those things that causes problems.

If it worked well before, I would assume that Borland''s STL puts a pointer to the string as its first element (i.e. you got lucky). VC++ uses Dinkumware''s STL, which apparently does not.

You don''t need it anyway. Use the c_str() member function to get a pointer to a char array from a string.

Don''t listen to me. I''ve had too much coffee.