[C++-sig] Re: Boost.Python scope

[Raoul Gough]

"Ralf W. Grosse-Kunstleve" <rwgk at yahoo.com> writes:

> --- Raoul Gough <RaoulGough at yahoo.co.uk> wrote:
> Congratulations to the VC6/7 workarounds for the indexing suite!

Wow - I haven't even announced the fixes yet! I guess you saw the CVS
checkin (BTW, it *is* all fixed, as far as I can tell). The MSVC6
supplied <deque> breaks the test code, but there is a free update for
this available from Dinkumware (will be documented when I check in the
html update). I tested this by installing the updated <deque> header
locally and altering the include path order to find the new version.

>
>> Sounds interesting.
>
> Here are the main links:
>
> http://mail.python.org/pipermail/c++-sig/2002-June/001503.html
> http://mail.python.org/pipermail/c++-sig/2002-July/001672.html
>
> http://cvs.sourceforge.net/viewcvs.py/boost/boost/boost/python/Attic/py_interface.hpp?rev=1.2&view=auto
> http://cvs.sourceforge.net/viewcvs.py/boost/boost/libs/python/src/Attic/py_interface.cpp?rev=1.2&view=auto
>
> For easy viewing, the attachment of the last link unpacked:
>
> http://cci.lbl.gov/~rwgk/boost_python/api_project/py_api_gen.py

Scary. Maybe too scary for me.

>
>> I assume you had some kind of C function declaration parser to drive it?
>
> Great idea; but to the best of my knowledge nobody was that ambitious so far.

It looks like David Hawkes did quite a bit of work on this. Judging by
the approach he took, parsing the C declarations wouldn't be
sufficient by itself, since he uses all kinds of decorations in the
API_List like "{new,err=NULL}" for return conventions and
"{decl=import_import}" for function renaming I guess.

>
>> I'd be tempted to take a look, but I should probably be focusing on
>> getting some gainful (i.e. paid) employment instead.
>
> Good luck! I am sure your work on the indexing suite makes for some good
> advertising.

Thanks. I'm sure I'll find something.

>
>> How much work do you think was left in it?
>
> Hard to tell. It also depends on how ambitious you are.

Well, put it another way - what, if anything, was wrong with what Dave
Hawkes had already done? Did it get dropped simply because the hard
(i.e. interesting) work was already done?

-- 
Raoul Gough.
(setq dabbrev-case-fold-search nil)