Tony Dietrich td@transoft.demon.co.uk wrote:
The problem with any standard like LSB is that everyone has their own belief as to what its essential to include in the standard, with the result that the final document is a bloated list of everyones' favorite widget.
This need not happen, though. In some cases, the standards group have decided to make the first standard a maximal subset of common features, as far as possible, and then layer on top of that over time, as consensus is reached. That doesn't seem to have happened with LSB for some reason, but it has happened with some popular recent programming languages. I know that Scheme is developing this way, with the IEEE and RnRS specifications being built upon by the SRFI process. I think Python's PEPs and Ruby's RERs work in a similar way, but those languages are specified by one implementation.