Avogadro Status

I accidentally sent this just to Marcus earlier.

It basically allows us to record interaction with the GUI and replay it in a unit test. As far as I know it only uses Qt (no additional dependencies) and is BSD licensed.

This would be invaluable. I assume that it also works cross-platform?

Geoff - I notice your dashboard fails as it is compiling Avogadro 64 bit, and glew is 32 bit. You could just turn off glew, and I could help you with a CTest script to ensure it updates properly - you can see the script I am using by clicking on the Note icon in the submission I made.

I seem to have another issue which cropped up yesterday, that on the Mac QtOpenGL is no longer included as a requirement, so some of the engines fail to link. I will try to chase that down today.

Anyway, if Kitware is willing to host our Gerrit server, I’m happy to point review.avogadro.openmolecules.net at anything we want.

The SBIR funding I applied for is through, and I will be working with Albert from the Gordon Group (GAMESS) on Avogadro, electronic structure, databases, cluster/HPC submission integration and other related tools. This means I get around 6 months where I can spend a fair portion of my days working on Avogadro, and if things go well this could turn into a few years.

This is great. I think one issue which has slowed development is that we don’t have a core developer working actively on Avogadro. So your funding obviously can make things move faster.

At the workshop at Cornell, there was a lot of interest in Avogadro and Open Babel. I’ll summarize this in a separate e-mail later today, but suffice to say that the solid-state community (e.g., Abinit) is definitely interested in Avogadro and would like to contribute. An Abinit developer is already working on an input generator for Git master.

-Geoff

Let me replay my response (which I didn’t notice had the list removed) for the list too.

On Nov 21, 2010, at 4:02 PM, Geoffrey Hutchison wrote:

I accidentally sent this just to Marcus earlier.

It basically allows us to record interaction with the GUI and replay it in a unit test. As far as I know it only uses Qt (no additional dependencies) and is BSD licensed.

This would be invaluable. I assume that it also works cross-platform?

Of course. It is used by ParaView on its dashboards, and has been around for several years in daily use on Windows, Mac, Linux and I believe a few more odd ball systems too.

Geoff - I notice your dashboard fails as it is compiling Avogadro 64 bit, and glew is 32 bit. You could just turn off glew, and I could help you with a CTest script to ensure it updates properly - you can see the script I am using by clicking on the Note icon in the submission I made.

I seem to have another issue which cropped up yesterday, that on the Mac QtOpenGL is no longer included as a requirement, so some of the engines fail to link. I will try to chase that down today.

Anyway, if Kitware is willing to host our Gerrit server, I’m happy to point review.avogadro.openmolecules.net at anything we want.

They are, and it frees me up from doing upgrades/maintenance while giving us a more powerful server that can drive dashboard submissions for patches. I may not have been clear, but it is the typical Gerrit installation - multiple projects hosted on one instance. We can have an Avogadro group etc, but there will be some switch over.

The SBIR funding I applied for is through, and I will be working with Albert from the Gordon Group (GAMESS) on Avogadro, electronic structure, databases, cluster/HPC submission integration and other related tools. This means I get around 6 months where I can spend a fair portion of my days working on Avogadro, and if things go well this could turn into a few years.

This is great. I think one issue which has slowed development is that we don’t have a core developer working actively on Avogadro. So your funding obviously can make things move faster.

Yes, and I would be very pleased to act as that person, as you allowed me to when I was at Pitt. My son is also getting older, and I am starting to get a little more free time :wink:

At the workshop at Cornell, there was a lot of interest in Avogadro and Open Babel. I’ll summarize this in a separate e-mail later today, but suffice to say that the solid-state community (e.g., Abinit) is definitely interested in Avogadro and would like to contribute. An Abinit developer is already working on an input generator for Git master.

That is great news. I also failed to mention all the work I have been doing with Quixote. Jens Thomas is already pushing GAMESS-UK and DLPOLY2 code into Open Babel/Avogadro. I have been working with Peter on Molpro support too, and Peter on some general data structure, database interactivity etc. Carsten also got in touch this morning off list about some possible joint work (I will let him volunteer more if he wants to).

I wish I could have made it to the workshop myself, maybe next time. I should be able to make the ACS meeting in March, if others can make it this could also be a good place/time to talk about the future and see how far we have gotten.

Marcus

At the workshop at Cornell, there was a lot of interest in Avogadro and Open Babel. I’ll summarize this in a separate e-mail later today, but suffice to say that the solid-state community (e.g., Abinit) is definitely interested in Avogadro and would like to contribute. An Abinit developer is already working on an input generator for Git master.

I’d like to remind you about discussion concerning universal framework for input plugins
A lot of time could be saved


Regards,
Konstantin

On Nov 21, 2010, at 4:29 PM, Konstantin Tokarev wrote:

I’d like to remind you about discussion concerning universal framework for input plugins
A lot of time could be saved

I agree completely. I think we need to do more work here, or Avogadro 2.0 will have 60+ input generators. :frowning:

In many cases, the hard work of the input generator is the GUI itself. For example, Abinit is very different from Gaussian or GAMESS, so they’re spending a lot of time in Designer.

I’d like to think we can cut out a good part of the code using QML in Qt 4.7+. This would also have the benefit that the “universal” class could load a QML file to add a new generator, rather than having to create new C++ code all the time.

-Geoff

22.11.2010, 00:39, “Geoffrey Hutchison” geoff.hutchison@gmail.com:

On Nov 21, 2010, at 4:29 PM, Konstantin Tokarev wrote:

I’d like to remind you about discussion concerning universal framework for input plugins
A lot of time could be saved

I agree completely. I think we need to do more work here, or Avogadro 2.0 will have 60+ input generators. :frowning:

In many cases, the hard work of the input generator is the GUI itself. For example, Abinit is very different from Gaussian or GAMESS, so they’re spending a lot of time in Designer.

I’d like to think we can cut out a good part of the code using QML in Qt 4.7+. This would also have the benefit that the “universal” class could load a QML file to add a new generator, rather than having to create new C++ code all the time.

Why not just use QUiLoader?


Regards,
Konstantin

I agree completely. I think we need to do more work here, or Avogadro 2.0 will have 60+ input generators. :frowning:

This won’t be bad at all :slight_smile:
Bad things are
a) massive code duplication between this generators;
b) most of plugins are very basic and don’t really expose the capabilities of QC software
c) most of plugins have less capabilities (and usability) than Gamess plugin which is
the best in my opinion

We’ve discussed code generation from XML description of program using scripting languages.
Another option is interpretation of XML in Qt code at runtime for construction of typical UIs,
and loading of special UIs (I see them as additional “leafs” on Gamess-like Advanced tab)
using QUiLoader. QML support would be great, but I don’t think it’s the best tool for our
"boring" desktop-oriented GUIs :slight_smile:


Regards,
Konstantin