Re: [boost] Preview 2 of a template Geometry Library

From: Fernando Cacciola (fernando.cacciola_at_[hidden])
Date: 2008-03-27 18:20:20

Hi Arash,

> Fernando Cacciola wrote:
>>
>> P.S: We (the CGAL developers) are well aware of a couple of pending
>> improvements to ease the user experience, but I get the feeling that some
>> users just don't understand the inherent complexity that comes from our
>> fundamental goal of robustness.
>
> Some users do, but don't see what it has to be soo apparent. Think of
> it in this manner:
>
> There is a street where every shop is a restaurant, the restaurants
> are ordered from say a McDonalds/Fish'n'Chip to some high upscale
> "we'll scratch you're butt if it needs scratching" style restaurant.
>
> Now from a user POV I (and most other users I know )want to always
> start from the McDonald's end of the street regardless and slowly as
> I determine what I want to eat gradually walk down the street to the
> other end. Some days I'd be content with McDonalds other days I may
> prefer something more edible/healthy and maybe once in a while I will
> go to the upscale place at the end of the street.
>
Indeed, I undestand the parabola (or is it a metaphor) and I totally agree 
with the concept.

Coming up with the simplest deal for the simplest needs and the real thing 
for the real needs within the same library is the holly grail of framework 
design. But granted, this should a stated goal, even if really hard to meet.

> I don't have any exact evidence but from my experiences both in
> Academia and industry "high end robustness" in problems requiring
> computational geometry is the exception not the norm

Unfrotunately, yes.

> I'm not saying
> it is not important and should not be provided just that it would make
> more sense from a user experience/marketability pov that you provide a
> cleaner simpler interface.
>
Absolutely... FWIW in CGAL we are trying to move towards a cleaner interface 
for simple needs.

> I digress - this is a BOOST ML and BOOST.Geometry is the topic at
> hand not issues relating to CGAL.
>
Right, but some degree of constructive cricticism of CGAL is right on topic 
as food for thought for the new library. My request intended to set a case 
which could be used to distinguish simple from simplistic, as, at the end of 
the day, I choose a complex but powerful library over a simplistic one that 
pretended to be simple.

Best

-- 
Fernando Cacciola
SciSoft
http://fcacciola.50webs.com
http://groups.google.com/group/cppba