Excerpts from Bram Moolenaar's message of Sat Jun 08 12:56:38 +0200 2013: > C++ is incredibly complex. I didn't ask about introducing all kinds of odities. I know that doing C++ proberly takes 2 years or more studying C++.
Now you can shoot yourself into your feet using any language. I just want "sane collections" - and C++ seems to provide them. That's why I consider using a C++ subset to be an option. With zimbu I see one issue: Is it possible to rewirte "one collection list" only which will be found in many files? I guess you have to rewrite the whole .c file using zimbu then. This would also force everybody to learn zimbu. While I may agree that its "yet another nice language" - eg like gnome's vala - I'm not sure its the easiest way to go. It would mean rewriting vim using zimbu partially. I mainly care about introducing some threading. Eg node has also written a 'libvu' abstracting over some issues. I'm not primarly a C programmer (even though I know what pointers are, and that I left Delphi long time ago because there were untyped lists only) Also libnih (from upstart) provides some simple list like iterators, and much more - I think even dbus integration. I don't want to read/write/review/debug list code. We have 2013. That's why I wonder what is best for the project. Eg for the buffer list case - who cares whether you're using a plain array, or a linked list? Who has more than 2000 items? And till 2000 items a simple array (and increasing array size by 50% always) should be equally fast - if not faster, because less mallocs have to be done. I also know about the yzis project (vim as component rewritten in C++). But its community is not that strong - and they support only lua (which would not be that bad). While I made it compile the project looked almost dead to me. Marc Weber _______________________________________________ nix-dev mailing list nix-dev@lists.science.uu.nl http://lists.science.uu.nl/mailman/listinfo/nix-dev
