Hi all,
I’ve made some modifications to the ball and stick engine to draw correctly
bonds that cross periodic boundaries. However, I cannot push the commit
because gerrit rejects my ssh connection.
I created a gerrit account and generated an ssh public key. Yet, I get the
error
$ git gerrit-push
Permission denied (publickey).
fatal: The remote end hung up unexpectedly
The public key I pasted to my gerrit account matches my id_rsa.pub file for
sure. I went through all the troubleshooting instructions to no avail.
I don’t know how to solve this problem. Therefore I’m sending the patch here
as an attachment.
Best,
Tuukka Verho
Hi Tuukka Verho,
I’m not sure why gerrit isn’t behaving for you – Marcus may have some
ideas, as he is most familiar with Gerrit.
I’ve made some modifications to the ball and stick engine to draw correctly
bonds that cross periodic boundaries.
This interests me quite a bit – I’ve been working on adding this
feature as well, although not much lately. I have a partial
implementation here:
https://github.com/dlonie/avogadro/commit/1457bedd6ae9e932f758d7c410541de3142475d3
It’s a bit buggy still, but I hope to get it cleaned up soon.
I tested out your patch, and it needs some work before it can be
merged in – I kept getting renderings like the attached, where the
bonds don’t line up with the cell boundary.
Dave
On Fri, Apr 13, 2012 at 12:59 PM, Tuukka Verho tuukka.verho@aalto.fi wrote:
Hi all,
I’ve made some modifications to the ball and stick engine to draw correctly
bonds that cross periodic boundaries. However, I cannot push the commit
because gerrit rejects my ssh connection.
I created a gerrit account and generated an ssh public key. Yet, I get the
error
$ git gerrit-push
Permission denied (publickey).
fatal: The remote end hung up unexpectedly
The public key I pasted to my gerrit account matches my id_rsa.pub file for
sure. I went through all the troubleshooting instructions to no avail.
Gerrit is working fine, you need to ensure you get both the correct
public key and the correct user name. This is not like github where
everyone uses git@, each user has a username. You can try connecting
directly using,
ssh user@review.source.kitware.com
If it recognizes you it will have the name you supplied. You can use
-vv to get more detail on the keys your SSH client is trying.
Thanks,
Marcus
This interests me quite a bit – I’ve been working on adding this
feature as well, although not much lately. I have a partial
implementation here:
https://github.com/dlonie/avogadro/commit/1457bedd6ae9e932f758d7c410541de314
2475d3
It’s a bit buggy still, but I hope to get it cleaned up soon.
I tested out your patch, and it needs some work before it can be
merged in – I kept getting renderings like the attached, where the
bonds don’t line up with the cell boundary.
Dave
Yes the stumps are not clipped exactly at the boundary, instead their length
is always just half the bond length. This was a conscious decision as I didn’t
see the need to have them clipped exactly “right”. Your approach is more
sophisticated but it’s also a lot more code.
I’m dealing mostly with amorphous polymer systems with lots of atoms. In such
cases it doesn’t make much difference how the stumps are drawn – perhaps in
the case of covalently bonded crystals there are aesthetic factors that should
be taken into account.
Just let me know what you think.
Thanks Markus, I was able to connect directly as you indicated. However I
won’t push until it is settled whether David’s approach is preferable or not.
On Friday, April 13, 2012 01:27:23 PM Marcus D. Hanwell wrote:
On Fri, Apr 13, 2012 at 12:59 PM, Tuukka Verho tuukka.verho@aalto.fi wrote:
Hi all,
I’ve made some modifications to the ball and stick engine to draw
correctly
bonds that cross periodic boundaries. However, I cannot push the commit
because gerrit rejects my ssh connection.
I created a gerrit account and generated an ssh public key. Yet, I get the
error
$ git gerrit-push
Permission denied (publickey).
fatal: The remote end hung up unexpectedly
The public key I pasted to my gerrit account matches my id_rsa.pub file
for
sure. I went through all the troubleshooting instructions to no avail.
Gerrit is working fine, you need to ensure you get both the correct
public key and the correct user name. This is not like github where
everyone uses git@, each user has a username. You can try connecting
directly using,
ssh user@review.source.kitware.com
If it recognizes you it will have the name you supplied. You can use
-vv to get more detail on the keys your SSH client is trying.
Thanks,
Marcus
On Fri, Apr 13, 2012 at 1:38 PM, Tuukka Verho tuukka.verho@aalto.fi wrote:
Thanks Markus, I was able to connect directly as you indicated. However I
won’t push until it is settled whether David’s approach is preferable or not.
I prefer my approach, but I’m obviously a bit biased 
I work with
smaller molecular crystals (Z=1-8) and the renderings look rather
strange when the bonds don’t line up with the cell.
That said, my code is still very rough and crashy. I’ll try to clean
it up this weekend and see if I can get it ready to merge.
Thoughts?
Dave
David Lonie kirjoitti 14.4.2012 kello 16.22:
On Fri, Apr 13, 2012 at 1:38 PM, Tuukka Verho
tuukka.verho@aalto.fi wrote:
Thanks Markus, I was able to connect directly as you indicated.
However I
won’t push until it is settled whether David’s approach is
preferable or not.
I prefer my approach, but I’m obviously a bit biased I work with
smaller molecular crystals (Z=1-8) and the renderings look rather
strange when the bonds don’t line up with the cell.
That said, my code is still very rough and crashy. I’ll try to clean
it up this weekend and see if I can get it ready to merge.
Thoughts?
Dave
You’re probably right. However I think I’ve come up with a quite
simple way to make the bonds line up with the cell. I will probably
send another patch within few days.
I’m not going to touch the rendering code though, so the bonds wont be
cut along the cell face directions. You seem to have some code to do
that so maybe there will be an opportunity to combine parts of the
patches.
On Sat, Apr 14, 2012 at 2:32 PM, Tuukka Verho tuukka.verho@aalto.fi wrote:
You’re probably right. However I think I’ve come up with a quite
simple way to make the bonds line up with the cell. I will probably
send another patch within few days.
I’m not going to touch the rendering code though, so the bonds wont be
cut along the cell face directions. You seem to have some code to do
that so maybe there will be an opportunity to combine parts of the
patches.
Just FYI, I put in some time this morning cleaning up my patch:
http://github.com/dlonie/avogadro/tree/ENH_intercell_bonds
It seems to be working well. All that is left to add is multicylinder
rendering, and there is also some glitchyness when rendering
selections.
Dave
On Sat, Apr 14, 2012 at 2:32 PM, Tuukka Verho tuukka.verho@aalto.fi wrote:
You’re probably right. However I think I’ve come up with a quite
simple way to make the bonds line up with the cell. I will probably
send another patch within few days.
I’m not going to touch the rendering code though, so the bonds wont be
cut along the cell face directions. You seem to have some code to do
that so maybe there will be an opportunity to combine parts of the
patches.
Cool – feel free to take a look at what I’ve got and let me know if
you can spot any possible improvements:
https://github.com/dlonie/avogadro/tree/ENH_intercell_bonds
A bit more progress today, along with a screenshot. Multiple bond
intersections are now handled properly, and bond segments are drawn
between the appropriate boundaries, even when no atoms are directly
connected (see attached).
Still have to add a multicylinder implementation, color the bonds
correctly, and fix up GL picking before it’ll be completely ready.
Dave
On Sunday, April 15, 2012 09:51:09 AM David Lonie wrote:
Cool – feel free to take a look at what I’ve got and let me know if
you can spot any possible improvements:
https://github.com/dlonie/avogadro/tree/ENH_intercell_bonds
A bit more progress today, along with a screenshot. Multiple bond
intersections are now handled properly, and bond segments are drawn
between the appropriate boundaries, even when no atoms are directly
connected (see attached).
Still have to add a multicylinder implementation, color the bonds
correctly, and fix up GL picking before it’ll be completely ready.
Dave
Your screenshot looks quite nice, indeed multiple intersections are correctly
handled. Though I’m getting the feeling that the code is quite large, and the
support for coloring and higher order bonds still need to be added as you say.
While thinking of a more straightforward way to achieve the same result I
stumbled upon some example code to do boolean operations with opengl
http://www.hackchina.com/en/r/78488/csg.c__html
If the clipping could be done with a logical AND operation, multiple bonds
etc. would be handled without any extra effort. Also cases when the cylinder
intersects with not just one but two or three walls would be nicely handled.
I will experiment a bit and let you know if anything comes out of it.