"Daniel Schlyder" <daniel@bitblaze.com> wrote in message news:12236559906.20050330152342@bitblaze.com... | Thorsten Ottosen: | > hm..is this from a cvs snapshot? If so, please update. | | No, I'm using CVS HEAD. I just updated and recompiled with the same result. | I'm using MinGW GCC 3.4.2, BTW, sorry for not mentioning before. | | > what is the virtual_control::s__xxxxxxxxxx ? Why don't you write | > | > const std::set<std::string> foo = ...: | | s_unbindables_permanent is a static variable in a class named | virtual_control. ok. | > hm...don't call to adapter unless you are dealing with an adapter (queue, | > priority_queue, stack). | | Oh, I'm not much of a C++ expert, as I'm sure you understand. ;) Didn't know | the difference between a container and a container adapter. | | > Read about to_container() instead. | | Thanks. Using it, the code compiles. I see it's mentioned as being a | workaround. Is it really needed for GCC 3.4.2? Why then did it work before? yeah, good question. I will ahve to look into this...sometime... the regresion tests shown that list_of don't work too well with gcc. | (Not sure when I last worked on this code, might be as long as a month ago. | I can try to find out exactly when it stopped working (the CVS date, I mean) | if that would be useful.) I would be worth knowing if the old boost.1.32 code worked with your compiler. -Thorsten