I appreciate your taking the time to write up such a detailed email.
I understand the importance of OS X and put quite a bit of energy into
making it work. With that said, I am very unhappy about the situation
... and my unhappiness is directed at Apple.
The Apple 1.4 Java Virtual Machine has a number of bugs in it,
particularly related to graphics operations. I submitted several bug
reports to Apple back in Feb and Mar of this year, during the beta test
period. None of these bugs were fixed and I never received any response
I also have submitted several bugs to the Safari bug tracking system ...
again no response.
So, I am very frustrated and disappointed with Apple.
To get to your questions:
> We are accumulating a number of OS X issues. Jmol is hugely important
> to this OS, since MDL has chosen not to support it with Chime.
> a) I used to build an application from the Jars by dropping Jmol.jar
> into a program called Jar Bundler. It would then offer a drop down
> menu to select what it detected as the Main class. Clicking on
> "create application" would then create what is called an application
> bundle. double clicking that as with any other application would run
> Jmol. Starting around Jmol 4 (or maybe 5) Jar Bundler could not
> longer detect the Main class, and attempts to put it in by hand did
> not generate any application bundle that would work. Now the ONLY way
> to run Jmol on OS X is from the command line as
> java -jar jmol.jar
> which for students is pretty inconvenient. Can anyone recollect any
> reason why Jar bundler should now be failing to find the main class?
I am not familiar with this tool. But now that you have given me a
pointer I will track it down and try to figure out why it is not
> b) We are struggling with http://jmol.sourceforge.net/scripting.html
> on OS X browsers (IE, Safari. icab).
I have documented this issue under Platform dependencies, but apparently
have not made it clear enough.
and Java. This means that you cannot call a Java function from
Safari is based upon Konqueror. Konqueror introduced LiveConnect about 6
months ago. So I suspect (hope) that Apple will introduce this
functionality in their next major Safari release. In the meantime, there
is nothing we can do about it.
> None of the scripts seem to want to
> run. Turning debug on, we get
> "only objects may be called as functions" (from icab)
I am not familiar with icab, but I am sure that the problem is the
communications. However I have had difficulties with it as well.
> Thus in the section "Enter your RasmolChime script"
> the links below will appear in the box, but pressing
> execute script again produces the error above. I might add that the
> molecules do render (in iCab, IE) although Safari has some problems
> rendering the atoms (but not the bonds).
The sphere-rendering issue is a graphics bug in the apple JVM.
> Can you cast light on either of these two?
> PS We are working hard at converting as much old Chime/Rasmol script
> stuff as we can to Jmol,.
I would like to have feedback on scripting-related issues.
> PPS I should know the answer to this, but dont! The jmol.jar applet
> is about 500K in size. This is a bit too large to routinely ask
> people to
> download on "popular" pages where the user might have a modem etc. I
> think it should be possible to ask the user (or administrator) to
> place this class in a standard classpath somewhere, and have the
> system detect this without downloading the 500k applet each time.
> Have you looked into this?
The problem is browser & security related.
There are two types of applets ... trusted & untrusted.
An untrusted applet cannot access resources on the local machine (with
the exception of the core java classes).
A trusted applet has full access to the local hard drive and could load
classes off of the classpath.
The advantage of an untrusted applet is that it can be loaded in any web
page without requiring any authorization from the user. That is, there
is no install and there is no dialog box which pops up to ask the user
if they want to install the applet.
Jmol is an untrusted applet for this reason ... I didn't want to put up
a dialog box which might scare away end users. At least, that was my
Once the applet is downloaded it *should* stay in the browser cache. So
if a user goes back the following day then it should not download the
If you think that we need a trusted applet then we can discuss it.
Another advantage of a trusted applet is that it could access files
stored on the local machine ... or on a remote web server. That means
that you could build pages which referenced molecular model data files
stored on someone else's web server.
Finally, some additional comments about bugs on Apple.
I am in the process of completely rewriting the graphics engine to give
us improved performance and true 3D capabilities. The entire image will
be constructed in an array in memory. Then that complete image will be
blitted to the screen in one operation. That should effectively
eliminate all platform-specific graphics issues. This should effectively
eliminate most of the
Apple graphics bugs. (Although they still have some windowing-related
Give me your feedback, let me know what you think & let's keep the dialog
> Henry Rzepa.
> +44 (0870) 132 3747 (eFax)
> http://www.ch.ic.ac.uk/rzepa/ Dept. Chemistry, Imperial College
> SW7 2AZ, UK.
> This SF.Net email sponsored by: Free pre-built ASP.NET sites including
> Data Reports, E-commerce, Portals, and Forums are available now.
> Download today and enter to win an XBOX or Visual Studio .NET.
> Jmol-users mailing list