Re: [Cdk-devel] Rendering issues

[Nina J. <jel...@gm...>]

On 24 October 2010 23:37, gilleain torrance <gil...@gm...>wrote:

> On Sun, Oct 24, 2010 at 8:53 PM, Nina Jeliazkova
> <jel...@gm...> wrote:
>
> >>Na+) are just text, and the bounds may not be computed properly,
> >>sorry. Text is really a pain.
> >
> > Hm, couldn't it be treated as a single atom, not just text ?
> Well, the difficulty comes when the bounding rectangle is calculated
> on atom positions (which are points). For a single atom, this is a
> single point, which is - according to Euclid - "that which has no
> part". In other words, no width or height and a bounds of (x,y,0,0).
> Text, on the other hand, has a definite size.
>
> > Is it possible to use the same "cleanup" method outside of JCP, for
> example
> > for rendering images like this
> >
> >
> >
> http://apps.ideaconsult.net:8080/ambit2/depict?search=[O-]C1%3DCC%28[N%2B]%28[O-]%29%3DO%29%3DCC%3DC1OC.[Na%2B]
> >
> http://apps.ideaconsult.net:8080/ambit2/depict?search=[NH4%2B].CP%28%3DO%29%28O%29CCC%28N%29C%28%3DO%29[O-]
> Thanks for the links, it's good to see how JCP rendering compares to
> other toolkits. Badly, it seems...
>

 Well, these two are not good examples I admit ...


> 1) Either the bonds are being drawn on top of the atom symbols, or the
> symbol background is not drawn.
>

Well, it was working fine in the previous setup (cdk 1.3.3 & JCP 3.0.1) but
today I've upgraded to cdk 1.3.6 and JCP 3.1.3 and something has changed
outside of my code ...


> 2) In the phosphyl-cysteine thing, the atom symbols are not centered
> on the atom positions.
>

Interestingly,   this doesn't have the same issue

http://apps.ideaconsult.net:8080/ambit2/depict?search=CP%28%3DO%29%28O%29CCC%28N%29C%28%3DO%29[O-]


> 3) Hydrogens are missing from the NH4+ ion and amine groups - that
> could be due to a preference setting I guess.
>

It's default preferences - could you give me a hint how to set preferences
in code, without touching JCP properties ?


> 4) The counter-ion text font is very different size to the larger
> structure.
>

Yep, this is because of me allocating equal size boxes for the two (or more
if any) parts and scaling each structure separately.  Is it possible to
force same scaling ?


>
> So you could use the ControllerHub without JCP, certainly. It's in the
> controlbasic module, and could be used without connecting up all the
> bits. That is, you wouldn't be interacting with the images that you
> produce by a servlet, but you could pass in null for references to
> things that aren't being used.
>
> Of course, it would be nicer if there were dedicated MoleculeSetLayout
> classes...
>

MoleculeSet is a slightly different thing IMHO,  just having a single
procedure to get an image of this kind of disconnected structures would save
lot of time.


Thanks for the quick reply this late!


Best regards,
Nina


>
> gilleain
>