Camelia, the Perl 6 bug

IRC log for #cdk, 2011-09-04

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
05:39 egonw joined #cdk
06:51 jbrefort joined #cdk
09:34 maclean joined #cdk
09:34 maclean morning
09:35 egonw hej maclean!
09:35 maclean hej egonw
09:35 egonw really... I was just away for half an hour from my laptop :)
09:36 maclean I solved cheminformatics yesterday, actually - it wasn't so hard...
09:36 egonw hehehe
09:36 egonw ah, so the ad was yours?
09:36 maclean Well, it took me all day!
09:37 maclean So, er, I had to advertise it :)
09:37 egonw yeah, monkey makes the world go round...
09:37 egonw umm... money... money
09:37 maclean I knew what you meant.
09:37 egonw :)
09:38 maclean What I have almost solved is generating simple graphs (and other combinatorial obj). Only missing 3% of the degree-three graphs on 12 vertices.
09:38 maclean *sigh* one day, one day..
09:39 egonw yeah, it's though stuff
09:39 egonw tough?
09:39 egonw yeah, it's difficult stuff
09:39 egonw (like English :)
09:41 maclean Well, there's this great quote in a paper about molgen that says something like "So we could generate millions of structures a second, but chemists still preferred the slower methods that were more flexible ... so we decided to listen to our customers" :D
09:46 maclean Anyway : thankyou egonw for the recent Cu/Pt/Mo/Sb/Gd - is that all of them?
09:48 egonw all I did yesterday
09:48 egonw I still have some left, as listed in the report
09:48 egonw that quote... makes you wonder what they did...
09:49 egonw what flexability?
09:49 egonw why was that flexability causing a significant slow down...
09:50 maclean Well, 'flexibility' in the sense of using more information, not conformational flex.
09:50 maclean Like, using NMR constraints, prescribed substructures, cycles/no cycles, etc.
09:50 egonw well, I get that...
09:50 egonw ah, ic...
09:50 egonw so, the still generate millions of structuers, but just few are meeting all requirements
09:51 egonw and generation millions of those per second is out of their capabilities?
09:51 maclean Well, they say basically that it's no use generating more structures than you can analyse.
09:51 egonw ah, ic
09:51 maclean You have to generate just those that you wanted, then analyse (or store) them.
09:52 egonw well, but that's the whole interesting thing here...
09:52 egonw a exhaustive structure generator is not the issue
09:52 egonw it's generating the right structures (like no duplicates, which is one they solved)
09:52 egonw but that also includes generating structures meeting certain criteria
09:53 egonw but still, they're the best around
09:53 egonw don't let me downgrade molgen
09:53 maclean They certainly have an impressive list of features.
09:53 egonw I always wondered if 5M from welcome trust would have bought them out
09:54 maclean Heh, I've no idea. The Welcome trust are ...er... welcome to throw money my way if they want.
09:54 egonw yeah, if only that chinese post-doc from Chris' group hung on a bit longer (addressing the problems we found), he could have been famous
09:55 maclean Of course, money wouldn't necessarily mean any of us will be able to understand double coset representatives, or homomorphisms of graph reductions.
09:56 maclean (or whatever)
09:56 egonw no, that's not something you learn in an afternoon
09:57 maclean I have books on this kind of thing - but that didn't help as much as I thought it would :(
09:59 maclean Anyway, I'm off to buy a desk. See you later.
09:59 maclean (metaphorically)
09:59 egonw no, this is where interdisciplenary research is needed
09:59 egonw ok, cheers
10:00 maclean Well, if you know any mathematicians willing to help...
10:00 maclean bye
10:00 egonw (you need to talk to a math guy interested in cheminfo)
13:47 CIA-38 cdk: maclean cdk-1.4.x * rc562129 / (3 files in 3 dirs): Mg final - http://git.io/APKM-w
13:47 CIA-38 cdk: Egon Willighagen cdk-1.4.x * r81ce6e5 / src/main/org/openscience/cdk/at​omtype/CDKAtomTypeMatcher.java : Hooked in Mg atom type detection - http://git.io/1XY1KA
14:06 CIA-38 cdk: maclean cdk-1.4.x * rae54e0c / (3 files in 3 dirs):
14:06 CIA-38 cdk: Tl final
14:06 CIA-38 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://git.io/SRxXLg
14:14 CIA-38 cdk: maclean cdk-1.4.x * r6207b4f / (3 files in 3 dirs):
14:14 CIA-38 cdk: Pb final
14:14 CIA-38 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://git.io/FQETvw
14:40 jonalv joined #cdk
14:42 egonw hi jonalv
14:42 jonalv hi egonw
14:42 egonw just adding another interesting paper from Bajorath's group on FPs
14:42 jonalv oh
14:42 jonalv link?
14:42 egonw How Do 2D Fingerprints Detect Structurally Diverse Active Compounds? Revealing Compound Subset-Specific Fingerprint Features through Systematic Selection
14:42 egonw http://www.citeulike.org/u​ser/egonw/article/9637919
14:44 jonalv egonw: ah, sounds familiar
14:44 jonalv I think I got that from someone
15:02 jonalv_ joined #cdk
15:03 jonalv_ egonw: http://pastebin.com/psGPMX4m
15:03 egonw hahahaha
15:03 egonw well, you have tom compile things first
15:03 egonw ant clean dist-all test-dist-all
15:04 jonalv_ /home/jonathan/Work/cdk/build.xml:265: Unable to find a javac compiler;
15:04 jonalv_ com.sun.tools.javac.Main is not on the classpath.
15:04 jonalv_ Perhaps JAVA_HOME does not point to the JDK.
15:04 jonalv_ It is currently set to "/usr/lib/jvm/java-6-sun-1.6.0.26/jre"
15:04 jonalv_ FAIL
15:04 egonw :)
15:04 jonalv_ I think I give up
15:05 egonw sudo apt-get install openjdk-6-jdk
15:05 egonw yeah, install Debian GNU/Linux in a VM on OS/X
15:05 jonalv_ this is a pc
15:05 jonalv_ but yea
15:05 egonw yeah, install Debian GNU/Linux in a VM on that pc
15:05 jonalv_ hm, and run Windows?
15:05 jonalv_ it's an idea
15:06 jonalv_ means I could run my photocollection program and Eve at the same time :)
15:06 egonw no, I would never suggest to run Windows
15:06 egonw I'm not evil
15:06 jonalv_ oh you mean I should run debian in a VM under ubuntu?
15:06 jonalv_ but why?
15:07 jonalv_ egonw: do I need to run that command before I can test run cdk code in Eclipse as well?
15:07 egonw no, only ant clean runDoclet then
15:08 jonalv_ egonw: is there a documentation for this that  Ican look at next time?
15:08 jonalv_ so far I have only found the git introduction on the cdk web page
15:09 egonw I'd have to write some stuff
15:09 jonalv_ it still can't find a javac compiler
15:09 egonw we used to have a developer's guide
15:09 egonw should still be around
15:09 egonw sudo apt-get install openjdk-6-jdk
15:09 egonw try that
15:09 egonw that will give a javac
15:09 jonalv_ yea that's what I did
15:10 egonw check: ant info
15:10 egonw and try this:
15:10 egonw ant update-alternatives  --config javac
15:10 egonw oops
15:10 egonw sudo update-alternatives  --config javac
15:10 jonalv_ There is only one alternative in link group javac: /usr/lib/jvm/java-6-openjdk/bin/javac
15:10 jonalv_ Nothing to configure.
15:11 egonw mmm
15:11 jonalv_ Unable to locate tools.jar. Expected to find it in /usr/lib/jvm/java-6-sun-1.6.0.26/lib/tools.jar
15:11 egonw not ant should be OK...
15:11 egonw it should not be looking there...
15:11 jonalv_ this is strange
15:12 jonalv_ I am guessing you ahve run the tests? Do they pass for you?
15:12 egonw dunno... don't even have your branch here yet
15:12 * jonalv_ tried on the mac instead
15:15 jonalv_ egonw: that branch is up now
15:15 jonalv_ on github
15:15 jonalv_ or rather that commit
15:17 egonw ok, will try... hang on
15:17 egonw git fetch jonalv :)
15:18 egonw git checkout -b 323-14x-klekotaRoth jonalv/KlekotaRoth
15:19 egonw no, it does not compile :)
15:19 egonw for a practical reason at least...
15:19 egonw I will make comment on GitHub now
15:20 jonalv_ it's frustrating when the tools aren't working. I couldn't even run the test in Eclipse
15:21 egonw true
15:24 egonw ok, left a few initial comments
15:24 egonw one about the missing @cdk.module in the FP impl
15:25 egonw causing the current build fail at my machine
15:28 jonalv_ oh
15:28 jonalv_ I thought I added that :(
15:28 jonalv_ was that only in the test class then
15:29 * jonalv_ needs to learn the cdk boilerplate
15:29 jonalv_ has anyone written anything about these things?
15:31 egonw well, this is what my groovy cheminfo book is about
15:31 egonw but I have so much to do, so things go slow
15:31 egonw the problem lies in keeping things update
15:31 egonw d
15:32 jonalv_ so it explains the cdk module system?
15:33 egonw no, not really
15:33 * jonalv_ still considers the module system to be "here be dragons" areas
15:34 jonalv_ but egonw, thanks for the comments. I will look at them soonish and ping you when fixed :)
15:34 egonw ok
15:35 egonw (dragons are good)
16:07 CIA-38 cdk: maclean cdk-1.4.x * r458d1aa / (3 files in 3 dirs):
16:07 CIA-38 cdk: Sr final
16:07 CIA-38 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://git.io/5p9jCA
16:32 egonw joined #cdk
17:03 CIA-38 cdk: maclean cdk-1.4.x * r8026156 / (3 files in 3 dirs): Ni final - http://git.io/vlmz-Q
17:03 CIA-38 cdk: Egon Willighagen cdk-1.4.x * r682d0a1 / src/main/org/openscience/cdk/at​omtype/CDKAtomTypeMatcher.java : Removed obsoleted code by Ni patch - http://git.io/VpL9iw
17:15 CIA-38 cdk: maclean cdk-1.4.x * ree1ac09 / (3 files in 3 dirs):
17:15 CIA-38 cdk: Ti final
17:15 CIA-38 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://git.io/ky54hw
18:00 sneumann joined #cdk
18:13 egonw hi sneumann
18:13 egonw http://pele.farmbio.uu.se/jenki​ns/job/Bioclipse.metabolomics/
18:18 egonw sneumann: in short... it compiles :)
18:19 egonw next station: see if it actually runs :)
18:19 egonw but not today
18:19 egonw bye
18:50 jonalv joined #cdk
19:23 sneumann_ joined #cdk
19:32 jbrefort joined #cdk

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary