freenode/#clasp - IRC Chatlog
Search
13:24:51
usha
@cracauer - could you let me know how to proceed building the docker image of cando with my system?
13:25:57
usha
i understand that we can write a .lisp program and use clisp to get the output... how do we do the same thing with clasp?
13:26:22
usha
could anybody send me an example of how to write a simple clasp program and get the output?
13:27:00
drmeister
Hello usha: I'm in France at the moment at a conference. I gave Martin everything I had for building every form of clasp.
13:27:29
drmeister
He's probably moving into our incubator space today and then should be looking at the build system to automate everything.
13:27:50
drmeister
You can write a .lisp program and use clasp to get the output just like with clisp.
13:29:20
drmeister
There are other, more sophisticated ways to write software in Common Lisp using the Slime development environment. But creating a .lisp file and loading it is the simplest way.
13:32:01
drmeister
Martin hasn't built a docker image yet - but he has all of the scripts that I put together integrating his suggestions. I think he's going to familiarize himself with them this week as he automates things.
13:40:14
usha
Thank you the test file that I had worked fine. I am getting familiarized with lisp so that I can help write the TI module as well as with tleap and cando comparison
13:58:57
drmeister
Maybe you could come to Temple soon for another visit - if you are tackling Lisp we can give you a strong push in the right direction.
14:01:03
drmeister
It is problematic to compare excluded atom lists - but pairwise nonbond terms are easy to compare.
14:14:25
shiho
It's ready for comparing tleap and cando topology files. I'm trying to read both and checking the results.
14:27:27
usha
Sure @drmeister I can make a temple visit... please let me know preferable weeks for the visit and I will schedule it accordingly
14:29:46
usha
if we can compare whether all the steps taken by tleap and cando are the same it would be great
17:26:29
cracauer
well, it is more like infrastructure work. Especially for sending chemical computation jobs to worker computers. But yes, I use Clasp for almost everything now.
17:29:20
cracauer
Oh and those pesky python and js packages that are needed for cl-jupyter. Why do they have 3 package managers each?
17:30:17
Shinmera
drmeister also hinted that you might be able to take a look at clasp's usocket backend, which is pretty broken at the moment.
17:32:17
Shinmera
Well, in this case broken to the point where server sockets don't work at all and lots of tests fail
17:32:18
cracauer
Trying to feed a contiguous build now so that usha and others can pick up a working tree with cl-jupyter and cando.
17:41:29
usha
@shiho - the topology.lisp is the one that has the save-amber-parm-format, that generates the parameter files right?
18:35:37
shiho
Now, I'm developing read-amber-parm-format which is topology file reader to compare tleap and cando.
19:29:01
usha
@Shiho, could you make the exponent e symbol to E like the tleap so that it is easy for me to compare - because diff command sees it as different
19:43:32
drmeister
The order of all of the terms can be different and still describe the same system.
19:45:31
usha
i am comparing /cando/src/lisp/leap/topology.lisp and /amber/AmberTools/src/leap/src/leap/utilio.c
19:46:33
drmeister
usha: In leap I think there is more intrinsic ordering of objects. Atoms, residues, molecules are kept in the order that they are loaded.
19:47:22
drmeister
Trusting intrinsic order can create some really nasty bugs - so I don't like it. In Cando I try to avoid intrinsic ordering of things as much as possible.
19:47:56
drmeister
If objects are to be ordered - then those object keep a sequence number and they are sorted before they are written out.
19:51:09
drmeister
The tricky part is the nonbond interaction - because they aren't stored as a pair list - it's much more efficient to represent nonbonded interactions using an excluded atom list - to store what interactions are not calculated.
19:51:39
drmeister
To compare nonbonds you need to convert that into the interactions that are calculated.
23:05:25
cracauer
drmeister: the `pip3' switch only said "python3" on that machine, then picked up python3.6 but later we use python3.7, which doesn't have the module.
23:09:12
cracauer
I found some packages that want 3.6 over 3.7. However, no action is required now since we can just have pip3 pick python3.7 specifically.
23:22:41
drmeister
I found some nasty problems with pip3 - if you encounter them we should discuss and sort them out.
23:24:01
drmeister
With pip3 version 9 I saw that pip3 would grab the wrong version of libraries when there were two libraries that needed the same dependency but needed different versions. Sometimes the older version would be used.
23:24:27
drmeister
I upgraded to pip3 version18 and then the pip3 script fails because they made some changes to the pip3 script.