>To quote Linus Torvalds: "As such, my most important function is to say NO! >to people.
And rightfully so, but this nay-saying must be done on a sound theoretical basis. As for the CGI example, this is an entire subsystem, which traditionally does not belong in a kernel and can be implemented on top of it. It is a totally different story when a full implementation of a protocol is turned down because another protocol does not have the same features. The implementation of the missing features cannot easily be an addon in 3rd party software. We are not speaking of strange patches here, but about implementing the work of many good people having designed a standard and succeeding to have it approved after a lot of work has been done. As for standards it is also well-known, that it is the implementation, that drives the use of the them and the benefits we get from them. Limited implementation = lots of proprietary kludges which must be handled. - Jørgen Thomsen
