[Jmol-developers] Sascha's progress report

[Miguel <mi...@jm...>]

sascha wrote:
> I've made a lot of progress in the last few hours; this description
> definitely helps.

good

=5Bsnip=5D
>>Write a small test program that instantiates a SmarterAdapter. Then rea=
d
>>in a file using it.
>>
>>You do not need to store the contents anyplace, but you need to iterate=

>>through all the atoms and all the bonds.
>>
>>
>
> Yes=21 I ported this and I can now read in a Mol file. There's a bit of=
 a
> hiccup in that the .NET equivealent of BufferedReader doesn't have a
> =22reset=22 method, but that's easily worked around.

very good.

>>>Control flow:
>>>I'm rather confused by the  rendering pipeline -- Graphics3D, Viewer,
>>>FrameRenderer, Frame, Platform3D -- so many related classes,
>>>
>>>
>>
>>Graphics3D and Platform3D provide rendering primitives. You have
>>demonstrated that part working, so don't worry about them anny more.
>>
>>
>
> I just needed to step through the code to figure out the control flow.
> It also turned out that the Java->C=23 converter messes up the conversi=
on
> of virtual methods, so that the base class virtual method gets called
> unless the subclass method is marked =22override.=22 Crazy, I thought
> virtual methods *meant* that the subclasses version would be called.

I agree ... that seems quite strange.

>>The FrameRenderer's job is to render the Frame. It does this by
>> delegating
>>to BallsRenderer, SticksRenderer, RibbonsRenderer, etc.
>>
>>
> I'm almost there=21 I'm pretending that everything uses a BallsRenderer=

> for now.

OK.

You can easily disable all of the other shapes by commenting them out in
JmolConstants.java & Eval.java

Be careful when working with those classes. They are dynamically loaded s=
o
there is almost no compile-time checking going on. Any errors will show u=
p
at run-time.

=5Bsnip=5D
> Okay -- now that I'm getting a grip on the SmarterJmolAdapter, I think
> I'll be able to move things around so I can create the model
> programmatically.

good. that is the right way to do it.

> Ah, but they do, but not how you think --  if I can't get this Jmol por=
t
> to work, I have to go back to my lame built-from-scratch Direct3D
> molecule viewer.

But you would not have to roll back that far.

org.jmol.g3d offers a higher level of functionality than Direct3D ... at
least when rendering molecules.

Therefore, worst case you could still implement your own data model on to=
p
of g3d. You would get higher performance and a higher level of abstractio=
n
... spheres & cylinders.

=5Bsnip=5D
> Jmol is super-fantastic. It's worth the porting effort.

good. I'm glad that you are enthused.

> The various
> rendering styles are the main reason I wanted to use Jmol -- and the
> ability to read so many file types. The organic chemistry professor I
> work with said something like =22it would be SPECTACULAR if we could us=
e
> Jmol.=22
>
> The port is going to work, tonight I think=21 I've got the
> SmarterJmolAdapter reading in mol files, and I've got g3d, and I
> understand how the ModelManager and Frame work -- I'm moments away.
> Thanks for all the help.

Excellent.

Miguel