Avogadro 1.1.x and Linux Distributions

In the past, we said that Avogadro 1.1.x shouldn’t be packaged by Linux distributions, since the library ABI might break. I think fairly soon, we can guarantee it won’t break - at least until Avogadro 2.

Should we just go ahead and say “yes, please package this”? I think Avogadro 1.1.x is definitely better than 1.0.3, so I don’t see why we keep the old version. We can very easily push out a 1.1.2 or 1.1.3 release to address some of the rougher spots before Avogadro 2 firms up.

Thoughts?

-Geoff

On Mon, Feb 17, 2014 at 3:46 PM, Geoffrey Hutchison
geoff.hutchison@gmail.com wrote:

In the past, we said that Avogadro 1.1.x shouldn’t be packaged by Linux distributions, since the library ABI might break. I think fairly soon, we can guarantee it won’t break - at least until Avogadro 2.

Should we just go ahead and say “yes, please package this”? I think Avogadro 1.1.x is definitely better than 1.0.3, so I don’t see why we keep the old version. We can very easily push out a 1.1.2 or 1.1.3 release to address some of the rougher spots before Avogadro 2 firms up.

Thoughts?

If we want to guarantee no more big changes, why not just go ahead and
release/tag 1.2.0? I would be good with doing that, and I think it
would be nice to make it clear by releasing 1.2.0 and drawing a
proverbial line in the sand. There are a few changes pending that we
can get merged and make the release.

Marcus

If we want to guarantee no more big changes, why not just go ahead and
release/tag 1.2.0?

Good idea. Let’s see if we can get Eigen 3 support (in compatibility mode) soon. The website suggests that Eigen 3.3 (whenever that will be) will eliminate Eigen 2 support.

I’d also love to see a finished version of the polymer builder and Quantum Espresso extensions merged in and any obvious bug fixes.

Thanks,
-Geoff

Hi,

On Tue, Feb 18, 2014 at 11:35:48AM -0500, Geoffrey Hutchison wrote:

If we want to guarantee no more big changes, why not just go ahead and
release/tag 1.2.0?

Good idea. Let’s see if we can get Eigen 3 support (in compatibility
mode) soon. The website suggests that Eigen 3.3 (whenever that will
be) will eliminate Eigen 2 support.

I’d also love to see a finished version of the polymer builder and
Quantum Espresso extensions merged in and any obvious bug fixes.

Any progress on this? The Debian freeze is approaching in a couple of
months, so I it would be nice to include Avogadro-1.2.0 into the next
version.

Cheers,

Michael

Any progress on this? The Debian freeze is approaching in a couple of
months, so I it would be nice to include Avogadro-1.2.0 into the next
version.

I think we should probably fix as many bugs as possible and push out 1.1.x as 1.2, even though it’ll be missing a few features. Is there a date for the freeze? When do you want something?

The relative quiet is that Marcus came to visit Pittsburgh and we’ve been assembling a “hit list” to get as many things ready for an Avogadro v2 release. The new rendering architecture is a huge improvement, but day-to-day usability needs a bit of work.

Thanks,
-Geoff

I believe Avogadro 1.1.1 still has a bug regarding molecular dipole moments. I don’t have any examples right now, but at home I have performed some calculations on a few diatomic molecules and If I remember correctly, using the calculated bond length and partial charge provided by Avogadro, the reported dipole moment is too small. Unless the dipole moment is calculated in units other than

I saw your report, but I don’t think this is a bug.

Estimated dipole moments are corrected by a linear regression for Gasteiger charges to DFT dipole moments. I don’t remember the correction factor.

Moreover, there’s considerable error in predicted dipole moments even using QM methods. I don’t remember the error bars on the regression, but MOPAC has an error of ~ 1D:
http://openmopac.net/PM7_accuracy/PM7_accuracy.html

If you think there’s something going on with the calculation… please let me know.

-Geoff

Hi,

On Fri, Aug 29, 2014 at 04:51:49PM -0400, Geoffrey Hutchison wrote:

Any progress on this? The Debian freeze is approaching in a couple of
months, so I it would be nice to include Avogadro-1.2.0 into the next
version.

I think we should probably fix as many bugs as possible and push out
1.1.x as 1.2, even though it’ll be missing a few features. Is there a
date for the freeze? When do you want something?

The hard freeze is early November, but early October would be
appreciated to leave some time for integration etc.

The relative quiet is that Marcus came to visit Pittsburgh and we’ve
been assembling a “hit list” to get as many things ready for an
Avogadro v2 release. The new rendering architecture is a huge
improvement, but day-to-day usability needs a bit of work.

If avogadro2 is out by then, I can see about getting it packaged, but it
might be premature to replace avogadro1 with it for the Debian release.

Michael

On Fri, Aug 29, 2014 at 6:58 PM, Michael Banck mbanck@debian.org wrote:

Hi,

On Fri, Aug 29, 2014 at 04:51:49PM -0400, Geoffrey Hutchison wrote:

Any progress on this? The Debian freeze is approaching in a couple of
months, so I it would be nice to include Avogadro-1.2.0 into the next
version.

I think we should probably fix as many bugs as possible and push out
1.1.x as 1.2, even though it’ll be missing a few features. Is there a
date for the freeze? When do you want something?

The hard freeze is early November, but early October would be
appreciated to leave some time for integration etc.

Sounds like something we could work towards.

The relative quiet is that Marcus came to visit Pittsburgh and we’ve
been assembling a “hit list” to get as many things ready for an
Avogadro v2 release. The new rendering architecture is a huge
improvement, but day-to-day usability needs a bit of work.

If avogadro2 is out by then, I can see about getting it packaged, but it
might be premature to replace avogadro1 with it for the Debian release.

Avogadro 2 is totally compatible with being installed in the same
prefix as Avogadro 1, would it be reasonable to package it as as well
as Avogadro 1 rather than replacing? The executable is avogadro2, the
libraries have different names, share directory etc.

Just a thought, if not then I am fine with holding off but I think it
would be nice to consider. Thanks for pushing us on this - it would be
great to get what we can in before the freeze.

Marcus

Hi,

On Fri, Aug 29, 2014 at 07:04:25PM -0400, Marcus D. Hanwell wrote:

On Fri, Aug 29, 2014 at 6:58 PM, Michael Banck mbanck@debian.org wrote:

If avogadro2 is out by then, I can see about getting it packaged, but it
might be premature to replace avogadro1 with it for the Debian release.

Avogadro 2 is totally compatible with being installed in the same
prefix as Avogadro 1, would it be reasonable to package it as as well
as Avogadro 1 rather than replacing? The executable is avogadro2, the
libraries have different names, share directory etc.

Just a thought, if not then I am fine with holding off but I think it
would be nice to consider. Thanks for pushing us on this - it would be
great to get what we can in before the freeze.

That’s certainly possible, but new packages need to get through manual
archive review before they enter, and the last one I uploaded from
scratch needed a couple of months. I guess the backlog will be cleared
up, but who knows. So yeah, if avogadro2 is out by then it could also
be an alternative, if we manage to package it fast enough.

Michael

Thanks so much for the follow-up (off-list). I have a patch now. Somehow the conversion from electrons * Angstrom was deleted - probably as part of eliminating the empirical conversion we used previously.

http://review.source.kitware.com/#/c/16865/

This should definitely make it into v1.1.2.

Thanks very much,
-Geoff

Thank you all so much for a wonderful molecular modeling program. I look
forward to seeing Avogadro continually improve and for a greater segment of
the scientific community to use it!
Best regards,
Steve

On Saturday, August 30, 2014, Geoffrey Hutchison geoff.hutchison@gmail.com
wrote:

Thanks so much for the follow-up (off-list). I have a patch now. Somehow
the conversion from electrons * Angstrom was deleted - probably as part of
eliminating the empirical conversion we used previously.

http://review.source.kitware.com/#/c/16865/

This should definitely make it into v1.1.2.

Thanks very much,
-Geoff


Dr. Steven Petrovic
Professor of Chemistry
Southern Oregon University
1250 Siskiyou Blvd.
Ashland, OR 97520
(541) 552-6803