From: Thomas C Bishop (bishop_at_latech.edu)
Date: Tue Dec 22 2015 - 23:58:30 CST

My thought was as follows.

Raw DCD's by themselves can be rather large*, but stripping water,
skipping frames, or
"fake" DCD's that show some extracted dynamic (normal mode) beat the
heck out of
any movie I'll ever make.

I already have all my sims organized on disk and VMD scripts to
intelligently select, load and display them.
If urlload worked like the other mol loads then I could merely
"pre-pend" the necessary public_html
phrase and grant access to anyone, anywhere.

The advantage: they would literally see exactly what I see but still be
free to do things I never thought of.

In this regard a short vmd script could be submitted as supplemental
data for publication.

Cheers,
TOm

* note large is a relative term that continually changes... I remember
when the entire KS group shared a 4GB HD
and the day it overheated to the point that you couldn't touch it w/ out
getting burned.

On 12/22/2015 09:52 AM, John Stone wrote:
> Hi,
> On the road today, so just a quick addendum to Josh's note.
> After we taught VMD to automatically load PDBs over the web
> via their accession code, the 'urlload' feature didn't get much
> (any?) use anymore, so it never got a comparable command in the new
> APIs. Since Tcl has http download of any URL built-in, it seemed
> that in most cases people would use that rather than something like
> urlload. That being said, if people have some smashingly useful or
> exciting reason to bring the urlload feature into the new mol commands,
> I would certainly consider it.
>
> Cheers,
> John
>
>
> On Fri, Dec 18, 2015 at 06:00:02PM -0600, Josh Vermaas wrote:
>> Hi Tom,
>>
>> The mol urlload command looks like its on a totally different codepath
>> that is hardcoded to not append to a file. In cmd_mol.C (line 492), you
>> have this line:
>> int molid = app->molecule_load(-1, localfile, argv[2], &spec);
>> The -1 is what is killing you, as that tells VMD to create a new
>> molecule every time you call the command. Addfile does this instead,
>> where it had previously figured out which molid you wanted to add the
>> file to (Line 935, same file):
>> molid = app->molecule_load(molid, argv[2], type, &spec);
>>
>> If you wanted this feature, it'd be a matter of creating a new legal
>> argument for type that addfile would know how to handle as a special
>> case, since url isn't a legal type as far as VMD is concerned... yet.
>> Something like:
>>
>> mol addfile type url http://stuff
>>
>> would need trigger the file fetch via vmd_mol_urlload (which is actually
>> in a tcl script! scripts/vmd/vmdinit.tcl is where the code is hiding)
>> and then detect the ACTUAL filetype based on the extension. John might
>> have a more elegant way of doing this. Now I'm curious though. When is
>> this more useful than downloading them to your own machine first? DCDs
>> get big quick, and I can't imagine that getting them from the internet
>> is faster than downloading them once to a local disk and then using the
>> usual addfile commands every time you rerun a script.
>>
>> -Josh Vermaas
>>
>> On 12/18/2015 04:41 PM, Thomas C Bishop wrote:
>>> Dear VMD,
>>> i figured out the webpdb and urlload commands... even for loading a
>>> dcd....slick!
>>>
>>> but how do I "mol add url" instead of just urlload?
>>>
>>>
>>> Seems that unless I add the dcd to an existing structure there are
>>> display problems and other information that's not in a dcd that I can
>>> only get with a mol load equivalent.
>>>
>>> Thanks in advance for telling me how to do this.
>>> Happy Hollidays.
>>> Tom
>>>

-- 
******************************
    Thomas C. Bishop
     Tel: 318-257-5209
     Fax: 318-257-3823
    www.latech.edu/~bishop
******************************