"Charles C" <nomail@nomail> wrote:
> Shay <Sha### [at] cccc> wrote:
> > Warp wrote:
> >
> > >
> > > Well, then let's remove every "convenient" thing there is in povray and
> > > make it as hard to use as possible.
> >
> > The point on which we disagree is the nature of convenience or
> > usability. The relationship between feature expansion and usability has
> > been argued a thousand times - there's no point in continuing to repeat
> > it. If you think a large expansion of the (already buggy) procedural
> > portion of POV SDL is the most convenient (to users) way to accomplish a
> > job like format conversion, then so be it. I think you're nuts. I say,
> > learn a little Perl (that's what it's there for), or even better,
> > hard-code format conversion into the POV source code. I guess that
> > sounds nutty to you.
> >
> >   -Shay
>
> I have to say I like the direction Warp wants SDL to go...  Most of my POV
> coding is in the procedural portion and there are a lot MORE things I'd
> like to be able to do directly in SDL. That's what makes POV-Ray useful to
> me as opposed to *having* to turn to external things.  The idea of making
> source SDL which is usable to anybody who has POV-Ray is appealing.  When a
> project starts to require multiple languages, the number of people who'll be
> able to use it starts to drop.
>
> My thinking and my hope for in an updated SDL are largely what Warp
> mentioned.  Specifically, more use of the dot operator (even adding .xy
> ..xyz etc to for vectors), user definable data structures again with more
> use of dot operators, more control, i/o for arbitrary binary files, and
> better string manipulation.
>
> One really specific thing I'd like:  The currently-'experimental' but often
> used feature, splines, I think would benefit from a new syntax - I think
> that splines
> should be treated as a basically/almost-normal array of 4d vectors.
> MySpline(Value) would return a 3d vector point along a spline like is done
> now.  MySpline[Value] would return a 4d vector control point of the given
> array-position in the form <x,y,z,t> where t is the float-position along
> the spline's length.  MySpline[Value].xyz would return a 3d version of the
> control point. Transformations would apply to all values on the 'spline
> array' except for the t values unless specifically called for by use of dot
> operators.
>
> Charles
>
> PS (Woohoo! I was able to log on today from home with my high latency
> connection!)

Well, I consider all these features as being 'low_level' and 'detailed
design'. If we say that the new SDL will be more or less object-oriented,
we involve most of those features you and others have mentionned.
Object-orientedness can be implemented in quite a heavy way (eg Java, Ada),
or more 'light'. But in any case we have the dot operator, structures,
modules + visibility rules, attributes & methods etc .. etc ...

My opinion is that we have to make higher-level and main orientations
choices now before anything else. Some people are capable of envisionning
high levels of a system, some can extract general features from a messy
list of collected requirements of all kinds (there are structured methods
for that), etc ...

It is important to me that we have to focus on 'WHAT' (and why) rather than
'HOW'.

And my feeling is that the new SDL should be object-oriented yet
user-friendly and inherit POV syntax for what POV is VERY efficient for:
describing objects and textures.


Bruno

Post a reply to this message