VMD-L Mailing List

From: Josh Vermaas (vermaas2_at_illinois.edu)
Date: Thu Jun 20 2013 - 16:13:25 CDT

Well this is very interesting. I can't reproduce the problem. :( I
tested the original script on a similar machine (Mac OS 10.6.8) in
addition to the two linux boxes earlier, and the code works fine on all
3, with no missing variables and the list is filled up. Do you perchance
have the 64-bit version of VMD installed? I don't know all the
specifics, but it is built against different TCL libraries, and may have
been broken that way. Without a clear way of reproducing the problem, my
recommendation is to just see if reinstalling VMD magically fixes things.
-Josh

On 6/19/13 4:49 PM, Alex Richards wrote:
> That was certainly informative, but how thorny is thornier? Here is
> what I get when I execute the following:
>
> >> proc hello_world {args} {puts "Hello, World"}
> >> hello_world
> Hello, World
> >> trace add variable ::vmd_pick_event write hello_world
> >> trace info variable ::vmd_pick_event
> {write hello_world}
> >> puts $::vmd_pick_event
> can't read "::vmd_pick_event": no such variable
> >> puts $::vmd_pick_atom
> can't read "::vmd_pick_atom": no such variable
>
> After setting the trace, picking atoms in the OpenGL Display causes
> nothing to occur in the TkConsole. It's interesting that the info
> command shows the attached trace even though the variable doesn't seem
> to exist anyways. What would be the next step in determining why this
> is so? I see that vmd_pick_event was added in version 1.8.7 and my
> version purports to be 1.9.1. I did not download and install it
> myself, however, so I may simply replace the installation and see if
> that works. Also, the machine I'm using has OSX 10.7.
>
> On 06/19/2013 02:29 PM, Josh Vermaas wrote:
>> Hi Alex,
>>
>> I've tested it on both machines I have ready access to, and I have no
>> specific ideas as to why it works on my machines but not yours. There
>> are a number of potential failure points. First off, you need to make
>> sure the trace is working. For that, I'd start with a much simpler
>> proc (puts "Hello World" or something like that) to make sure your
>> procedure is activating when you pick an atom. If the trace is
>> working, the most likely point of failure is not being able to find
>> the variables... This is somewhat thornier, and should be
>> investigated by explicitly calling your proc (somehow trace quashes
>> error messages). On my machine, the variable $vmd_pick_atom exists in
>> the global namespace, so "puts $::vmd_pick_atom" always works.
>>
>> Hope this helps!
>>
>> -Josh Vermaas
>>
>> On 06/18/2013 03:31 PM, Alex Richards wrote:
>>> Josh,
>>>
>>> That is exactly what I was looking for! Except... it doesn't quite
>>> seem to be working for me. Here's my process. In VMD TkConsole, enter
>>>
>>> set ilist [list ]
>>> proc add_to_list {args} {lappend ::ilist $::vmd_pick_atom ; puts
>>> $::vmd_pick_atom}
>>> trace add variable vmd_pick_event write add_to_list
>>>
>>> Then I take the mouse, enter Query Mode (although I believe any
>>> left-clicking counts as picking...), pick atoms, at which point I
>>> can see the feedback in the Terminal, but nothing in TkConsole where
>>> I would expect to see the output of that puts command. Then, in
>>> TkConsole, if I enter
>>>
>>> puts $ilist
>>>
>>> it just returns an empty list.
>>>
>>> Noting the syntax highlighting in the TkConsole, I see that
>>> [::]vmd_pick_atom and [::]vmd_pick_event do not become highlighted
>>> as though they were variable names, which I think I would expect
>>> them to be. From everything I've read on callbacks, trace, proc,
>>> global, upvar, etc., I see no reason why your code wouldn't work
>>> except for perhaps not being able to access the global variables
>>> vmd_pick_atom and ilist, but I would think the :: would take care of
>>> that. At any rate, I've used all the combinations of:: andglobal
>>> that I can think of, but nothing seems to be taking.
>>>
>>> Also, the output of
>>>
>>> trace info variable vmd_pick_event
>>>
>>> is
>>>
>>> {write add_to_list}
>>>
>>> like I would expect.
>>>
>>> Any ideas?
>>>
>>> Thank You,
>>> A- R-
>>>
>>> On 06/17/2013 05:37 PM, Josh Vermaas wrote:
>>>> Hi Alex,
>>>>
>>>> I think you'll be interested in the TCL callbacks
>>>> (http://www.ks.uiuc.edu/Research/vmd/current/ug/node156.html). You'll be
>>>> wanting to do something like this in the tkCon window (or in your
>>>> startup script):
>>>>
>>>> #Make an empty list
>>>> set ilist [list ]
>>>> #Define a new proc to add new indices to the list.
>>>> proc add_to_list {args} {
>>>> lappend ::ilist $::vmd_pick_atom
>>>> puts $::vmd_pick_atom
>>>> }
>>>> #Add a trace so that the proc is called whenever an atom is picked
>>>> trace add variable vmd_pick_event write add_to_list
>>>>
>>>> After picking a few atoms, you should see that $ilist now contains the
>>>> indices of the selected atoms.
>>>>
>>>> Happy Picking!
>>>> -Josh Vermaas
>>>>
>>>> On 06/17/2013 04:32 PM, Alex Richards wrote:
>>>>> Hi All,
>>>>>
>>>>> I see that I can "pick" an atom in the OpenGL Display and have some
>>>>> identifying information about it posted into the terminal window. I
>>>>> would like to be able to "pick" an atom in such manner and automatically
>>>>> have it or its index added to a list which I can use in scripting. Is
>>>>> there a way to do this that has already been established? I usually use
>>>>> the TkConsole and I know that the atom info gets output to the Terminal
>>>>> window, so I'm not sure if I can move data back and forth between the
>>>>> two anyways. Thank you.
>>>>>
>>>>> A- R-
>>>
>>
>