This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Patch: stl_vector.h and vector.tcc


Paolo Carlini <pcarlini@suse.de> writes:

| Hi Gaby, everyone
| 
| Gabriel Dos Reis wrote:
| 
| >Well, to make it clear, I do not believe it is hard to implement.
| > (I was hoping to get some spare time and give it a try.) It -might-
| > however take some fair amont of time to get it in
| > mainline...
| A copuple of very different comments (I'm finally reading in detail n1660):
| - From a practical point of view, moving from cpp_type_traits.h to
| #nospam + <type_traits> seems almost matter of search & replace: nice!
| (obviously, the same cannot be said for type_traits.h)
| - I'm rather surprised to see 'value' being a static const instead of
| an enum: definitely the wrong choice for these applications according
| to, f.i., Vandevoorde, $17.2. So, out of curiosity: is current gcc
| actually able to deal well with the former?

I don't have a copy of David's book with me right now, but as the
original author of cpp_type_traits.h, I did not put any 'value' there,
and even less put any static data member there.  I used enums.  So, if
it changes then it must be me not noticing.  
The case of <type_traits> is different -- the authors of <type_traits>
believe in const static data members for various technical reasons.
That is what the committee adopted. 

I think GCC can deal with most "resonable" uses.

-- Gaby


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]