From: Gary Powell (Gary.Powell_at_[hidden])
Date: 2000-10-17 14:47:18
If the discussion were about FORTRAN or Algo68 interoperability, I don't
think there would be an objection.
Boost is about standards, and what py_c++ does is make a stab at
standardizing the interface between these two languages. Just because we
don't have an advocate or programmer to write Fortran04_c++, Snowball_c++,
Lua_c++, Ruby_c++, .... doesn't mean we should reject py_c++. Sure it
appears we are favoring one scripting language over another by only having
one interface, but if you want your scripting language included, write your
own library! After three or four of these interface libraries are written we
could come up with a "standard" c++ to XXXX language pattern, which instead
of the py_c++ library being the standard, it becomes an implantation of the
standard pattern. Until that happens py_c++ shouldn't be rejected.
The review process is a massive code/doc/design review. Not many
programmers are comfortable with this level of scrutiny. Once the review is
done, we then claim that this is quality code.
As for the layered approach, that would be my ordering for approaching the
standards committee. First things first, essential and missing, really handy
and missing, really nice and missing.
So on accepting or rejecting py_c++, my criteria is first, is the
documentation good, is the code good (I looked at this and yes.), is the
interface to python and C++ good? Since I don't use python I'm not qualified
to judge the last, since I'm lazy I again don't have a vote on the
documentation. On the code, its good. It passes.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk