I think these are all good points. In order to create a Jmol app for iPads, one would need to completely port the project to Objective-C, no easy task. To publish in Apple's App Store also costs money (99$/year), is very, very complicated and requires constant attention. And regardless of whether such an app would exist, Jmol authors would still need to "recreate" their content to work in these apps; the applets on their websites would still not work. So we are focusing on HTML5/WebGL as the solution.
Most of the performance issues with the ChemDoodle Web Components and large proteins have been addressed, and the toolkit should never crash when parsing any sized PDB file at this point. However, when using the backend services, a PDB file may take a while to parse, and our Apache server automatically terminates connections after 60 seconds, resulting in an error. This is due to the fact that our Java backend is based on the ChemDoodle desktop API, and therefore has overhead for working with the GUI and for creating 2D graphics. We are currently in the process of separating this logic out, so that really intensive functions become much more efficient. I should note that our graphics are not limited to just proteins, the ChemDoodle Web Components library handles small molecules, macromolecules, crystals, spectra, and 2D as well as 3D (MOL/PDB/CIF/JCAMP/XYZ). We are also working on surfaces (an experimental API is already included). Maybe this is a good time to really push for an integration of Jmol functionality on our backend though. We are still working on an efficient marching cubes algorithm in Javascript, and until that is complete, we could use Jmol instead. We would just need to know how to call it and export the necessary information. I think with your help, Bob, we could have this in a few days. So hopefully, as the project develops, it will become an easy procedure to substitute ChemDoodle for Jmol as the frontend on platforms that don't support Java. Certainly, we are doing our best to work towards that goal. As with all things, support matters, so if anyone is interested in seeing this project succeed, please help us by giving us suggestions and advice and by mentioning the open source ChemDoodle Web Components to your colleagues and students. On an aside, we are working with eTextbook vendors, so our technologies will be accessible to authors on a number of platforms. So working with us is a good way to get the functionality you want on the devices that you want. -Kevin On Apr 2, 2012, at 11:37 AM, Robert Hanson wrote: > OK, there are really three threads here: > > 1) Jmol app for the iPad > 2) Jmol export of iPad-ready model twiddles that can be embedded in iBooks > 3) Jmol applet-equivalent for browsers running in Java-challenged platforms > > (1) was the discussion I was starting. But now I think that's off track. I > guess I don't know if anyone really is interested in that. > > (2) is easy. Otis showed how that is possible, and we could make it simpler. > A Collada exporter is somewhere on my TODO list. > > Craig, if I hear what you are saying, you're not so much interested in (1) or > (2). You are interested in a browser-based scriptable applet, (3). > ChemDoodle's experiment is ongoing, and so far it is proving to be > marginally effective, as far as I can tell. It is a hybrid of (2) and (3) -- > a relatively unscriptable twiddlable model for the browser (with interesting > metainformation available from a server). > > Regarding (3), I'd need to see real-time molecular surface creation, > calculation of MOs from wave functions, decent translucency, and effective > graphical manipulation of large systems before I would endorse HTML5 as the > way to go in general. So I'm hoping the ChemDoodle experiment will continue > and we will hear more good stuff from them. The idea of porting 7 Mb of Java > code to interpreted JavaScript source, however, is mind numbing. That's just > asking for the impossible, I think, in terms of practical performance. (Of > course, people said that about Jmol being written in Java, as well...) I'm > very interested in hearing more about the practical limitations of HTML5. I > don't know why ChemDoodle was crashing with simple rendering of large > proteins last time I tried it (last summer, I think), but I will remind those > reading this that there is a reason Jmol works so well, and it is mostly > because we do NOT use standard graphical toolkits. > > Is that more to the point? > > -- > Robert M. Hanson > Professor of Chemistry > St. Olaf College ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Jmol-users mailing list Jmol-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jmol-users
