Re: Revision 23104

2008-02-21 Thread Stefan Schimanski
See my other mail on the list about exactly this problem. The set is not the right containter, but vector isn't either... Let's wait for some conclusion before optimising for the trash here. Well, add 'no additional includes if they are not absolutely needed' to your set of requirements the

Re: Revision 23104

2008-02-21 Thread Andre Poenitz
On Fri, Feb 22, 2008 at 12:35:16AM +0100, Stefan Schimanski wrote: > > Am 22.02.2008 um 00:33 schrieb Andre Poenitz: > >> On Fri, Feb 22, 2008 at 12:27:15AM +0100, Stefan Schimanski wrote: >>> I return a const pointer now. The shared pointer is not really essential >>> here because the ownership is

Re: Revision 23104

2008-02-21 Thread Stefan Schimanski
Am 22.02.2008 um 00:33 schrieb Andre Poenitz: On Fri, Feb 22, 2008 at 12:27:15AM +0100, Stefan Schimanski wrote: I return a const pointer now. The shared pointer is not really essential here because the ownership is clear enough. Better. The deque does not have to appear in the header eit

Re: Revision 23104

2008-02-21 Thread Andre Poenitz
On Fri, Feb 22, 2008 at 12:27:15AM +0100, Stefan Schimanski wrote: > I return a const pointer now. The shared pointer is not really essential > here because the ownership is clear enough. Better. The deque does not have to appear in the header either, it's private static data anyway... And I ju

Re: Revision 23104

2008-02-21 Thread Stefan Schimanski
Am 22.02.2008 um 00:24 schrieb Andre Poenitz: On Fri, Feb 22, 2008 at 12:17:38AM +0100, Stefan Schimanski wrote: Am 22.02.2008 um 00:13 schrieb Andre Poenitz: Stefan, it might be a bit late in the game, but having #include that you added in revision 23104 in Inset.h is not acceptable.

Re: Revision 23104

2008-02-21 Thread Andre Poenitz
On Fri, Feb 22, 2008 at 12:17:38AM +0100, Stefan Schimanski wrote: > > Am 22.02.2008 um 00:13 schrieb Andre Poenitz: > >> >> Stefan, >> >> it might be a bit late in the game, but having #include >> >> that you added in revision 23104 in Inset.h is not acceptable. This adds >> more than 17000 line

Re: Revision 23104

2008-02-21 Thread Stefan Schimanski
Am 22.02.2008 um 00:13 schrieb Andre Poenitz: Stefan, it might be a bit late in the game, but having #include shared_ptr.hpp> that you added in revision 23104 in Inset.h is not acceptable. This adds more than 17000 lines to each compilation unit where it was not used before (and quite a bi