Re: [h5md-user] fix remaining imprecisions - particle position

[Felix Höfling]

Am 10.01.2014, 14:51 Uhr, schrieb Pierre de Buyl  
<address@hidden>:

> On Fri, Jan 10, 2014 at 02:17:59PM +0100, Felix Höfling wrote:
> > I was not aware that you're actually using the offset. I've got the
> > impression from the mailing list that this piece of information is
> > implementation-dependent and is typically not stored in data files.
> > You may revert the respective commit if you feel that the field
> > should be present.
> >
> > Concerning the bounding box: it is a promise made to, e.g.,
> > visualisation software to choose the initial viewport without
> > scanning the whole(!) trajectory. Particles outside the box
> > (violating the promise) could silently be dropped. If it turns out
> > that such a bounding box is not of interest for anybody please
> > remove it.
> >
> > I made these changes to the repository to reflect the state of the
> > discussion. Otherwise, the suggestions on the list often remain
> > without feedback and fade away as time goes by. I hope that we can
> > really finalise version 1.0 without further delay.
>
> Hi Felix,
>
> Thanks for your reply. I also hope that this is over soon :-)
>
> I hadn't the visualization issue in mind, but I am sure that  
> visualization
> software (VMD for instance) compute it on the fly when reading particles'
> coordinates. I am trying to check this.
>
> Also, the fact that it is proposed now pushes me to put into  
> "discussion".
>
> I am using "offset", indeed, but if it bring confusion to others I am ok  
> to let
> it go.
>
> Best,
>
> Pierre

Hi Pierre,

I've just had a quick chat with Olaf on the phone. We found that a  
separate module for visualisation would be useful as more information is  
needed, e.g., by VMD. Such a module would a good place to define the  
bounding box, which is then not an attribute to position any more, but a  
proper H5MD element. (Allowing for fixed and changing bounding boxes.)

Felix