Camelia, the Perl 6 bug

IRC log for #cdk, 2008-12-04

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

All times shown according to UTC.

Time Nick Message
03:55 slyrus_ joined #cdk
05:21 slyrus_ joined #cdk
06:23 Gpox joined #cdk
07:35 CIA-16 cdk: goglepox * r13453 /cdk/branches/jchempaint-primary/src/main/org/o​penscience/cdk/controller/DrawBondModule.java: added @cdk.module control
07:35 CIA-16 cdk: goglepox * r13454 /cdk/branches/jchempaint-primary/src/m​ain/org/openscience/cdk/renderer/gener​ators/ExternalHighlightGenerator.java:
07:35 CIA-16 cdk: Generator for drawing external selected part.
07:35 CIA-16 cdk: The factors 1.3 and 10 are not tested and might be horibly wrong
07:35 CIA-16 cdk: goglepox * r13455 /cdk/branches/jchempaint-primary/src/m​ain/org/openscience/cdk/renderer/gener​ators/ExternalHighlightGenerator.java: Changed from Renderer2DModel to RendererModel
08:10 egonw joined #cdk
08:35 jpansanel moin
08:35 zarah hi jpansanel
09:21 shk3 joined #cdk
09:23 jonalv joined #cdk
09:35 CIA-16 cdk: jonalv * r13456 /cdk/branches/jonalv-fingerprint/: Created branch for fingerprint optimization code testing
09:36 jbrefort joined #cdk
09:57 jonalv -names
10:02 CIA-16 cdk: jonalv * r13457 /cdk/branches/jonalv-fingerprint/src/​main/org/openscience/cdk/fingerprint/ (Fingerprinter.java GraphOnlyFingerprinter.java): [jonalv-fingerprint-branch] applied the patch
10:04 shk3 egonw: anything new/done about
10:04 shk3 "regressions, which I suspect are caused by the Element.atomicNumber = UNSET patch"
10:04 shk3 ?
10:04 egonw no
10:04 egonw I think the only remaining thing is that it confuses the morgan number tool
10:05 egonw which is really unstable :(
10:05 egonw per these kind of changes...
10:05 egonw shk3: see for example the regressions in the unique SMILES generation
10:05 shk3 is the fingerprint test still failing?
10:06 shk3 sorry, didn't fully understand what you meant
10:06 egonw http://pele.farmbio.uu.se/nightly-​1.2.x/test/result-fingerprint.html
10:06 zarah egonw's link is also http://tinyurl.com/6rc8oa
10:06 egonw nothing seems to be failing...
10:06 shk3 great
10:06 egonw what FP test is stilling failing for you?
10:06 jonalv are cdk tests allowed to use System.out?
10:06 shk3 no, none
10:06 egonw no
10:06 shk3 just wanted to check
10:07 egonw shk3: that's what we have Nightly for...
10:07 shk3 did somebody change the atomNumber thing so that it behaves as before?
10:10 shk3 anyway, no reason to worry
10:10 shk3 that's good to know
10:10 jonalv egonw: I am afraid it might be your optimization that is failing... :(
10:10 egonw no, still is NULL
10:10 jonalv egonw: I will keep looking...
10:11 egonw shk3: because the number of new fails was rather minimal, and actually fixed a few things too...
10:11 egonw jonalv: yeah, would have too... but I don't understand yet why...
10:12 jonalv egonw: I did an ugly speedup after you did your rewrite too. I thought it was that one fiurst...
10:12 egonw shk3: and except for a fail in builder3d (please check that one, and let me know what you think of it), was the unique SMILES are now different... just differently unique
10:12 jonalv egonw: I am not sure about anything yet though...
10:12 egonw do you know how to run the FP tests?
10:12 egonw we may need to add a few to make sure the patterns do not change...
10:12 jonalv egonw: yea or that is I am running it as any other test..
10:12 egonw or, we need to figure out why they change
10:20 jonalv egonw: if I remember right the speedup from your rewrite wasn't huge was it?
10:20 egonw test it
10:20 egonw summarize of GDocs
10:20 egonw do it like this...
10:20 egonw mom
10:20 jonalv I have no test that times the figerprinter... :(
10:20 egonw we'll make one...
10:20 egonw please drop by
10:21 * jonalv dropes by egonw...
10:21 jonalv |drops
10:24 egonw http://cdk.svn.sourceforge.net/v​iewvc/cdk/cdk-qa/trunk/projects/
10:24 zarah egonw's link is also http://tinyurl.com/6rwz7t
10:25 shk3 is rendering of atom numbers already working in new jcp?
10:30 egonw not that I am aware of
10:30 egonw should not be too difficult...
10:31 shk3 just wanted to check
10:32 shk3 I will file a bug report with respect to rendering
10:32 shk3 just that things are not forgotten
10:37 egonw ack
10:38 egonw strictly speaking, these should go as feature requests, not bugs
10:38 CIA-16 cdk: jonalv * r13458 /cdk-qa/trunk/projects/081204-0001/: Initial import.
10:40 CIA-16 cdk: jonalv * r13459 /cdk-qa/trunk/projects/081204-0001/ (8 files in 4 dirs): Initial import of a little java plugin project for measuring time of fingerprinter
10:47 shk3 well, thought about that, but decided to do all jcp stuff as bug reports
10:48 shk3 everything which is missing for feature completness compared to old jcp is a sort of bug
11:07 jonalv uhm guys, help: he method getFingerprint(IAtomContainer) in the type Fingerprinter is not applicable for the arguments (IChemModel)
11:07 jonalv what's a chemmodel?
11:10 jonalv egonw: ?
11:10 egonw @gg IChemModel
11:10 egonw @google IChemModel
11:10 zarah egonw: http://chem-bla-ics.blogspot.co​m/2007/07/cdk-data-model-1.html
11:11 gilleain joined #cdk
11:12 jonalv how do I get an atomcontainer from a chemmodel?
11:12 egonw ChemModelManipulator.getAllAtomContainers()
11:12 egonw ChemModelManipulator.getAll​AtomContainers(IChemModel)
11:15 CIA-16 cdk: gilleain * r13460 /cdk/branches/jchempaint-primary/src/main/org/o​penscience/cdk/renderer/LogicalSelection.java: Missing cdk.module statement.
11:33 gilleain shk3 : about fit to size...
11:33 gilleain http://gilleain.tumblr.com/post/62985872
11:33 zarah gilleain's link is also http://tinyurl.com/5ojojs
11:34 shk3 looks good
11:38 * gilleain now understands
11:38 egonw how is 'natural size' defined?
11:38 egonw doesn't really look those are angstrom lenght bonds to me
11:38 egonw :)
11:39 gilleain at the moment, it's just a constant scale factor.
11:39 gilleain in any graphical program, there are a few 'magic numbers'
11:40 gilleain in old jcp I think it was bondlength/bondwidth
11:40 gilleain but there was also a scale factor of 60.0
11:40 egonw :)
11:40 egonw that was for those high resolution 9nm screens :)
11:41 gilleain the problem is when you get to ratios that look nice
11:41 gilleain I found with what I used to draw that a ratio of width/height of 2/3 was best for linear graphs.
11:55 CIA-16 cdk: shk3 * r13461 /jchempaint/trunk/src/main/org/openscience/jch​empaint/applet/JChemPaintAbstractApplet.java: most of the applet parameters are handled properly
12:00 CIA-16 cdk: gilleain * r13462 /cdk/branches/jchempaint-primary/src/main/org/ope​nscience/cdk/renderer/IntermediateRenderer.java: Uses the margin from RendererModel.
12:17 egonw http://www.journal.chemistr​ycentral.com/content/2/1/24
12:17 zarah egonw's link is also http://tinyurl.com/59zf62
12:18 egonw paper discussing CDK functionality
12:18 gilleain "Despite their complementary features, using these toolkits in the same program is difficult as they are implemented in different languages.."
12:19 gilleain Answer : Perl.
12:19 egonw hahaha...
12:19 egonw no, you misread the title...
12:19 egonw Python
12:19 egonw oh, btw
12:19 egonw who said Java was not fast:
12:19 gilleain d'oh.
12:19 egonw time to iterature over a SDF file:
12:19 egonw RDKit (native C++): 100 secs
12:19 egonw OB (native C++): 32 secs
12:20 egonw CDK: 21 secs
12:20 gilleain cool
12:23 CIA-16 cdk: shk3 * r13463 /jchempaint/trunk/src/main/org/openscience/jch​empaint/applet/JChemPaintAbstractApplet.java: cleanup
12:34 Gpox gilleain: do you want to do a DrawRingModule? like the DrawBondModule i did yesterday
12:35 Gpox it ganna complicate the phantom-Atom / Bond stuff
12:35 jonalv joined #cdk
12:36 egonw jonalv: http://chem-bla-ics.blogspot.com/200​8/12/who-says-java-is-not-fast.html
12:36 zarah egonw's link is also http://tinyurl.com/6fn5dp
12:36 gilleain hmmm. so AddBond/DrawBond, AddRing/DrawRing will be alternatives?
12:37 Gpox Well i think DrawBond is how you want to do it, it is all about the feedback while drawing
12:37 gilleain not necessarily.
12:38 gilleain click-sprout is fairly clear, and a fast way to draw.
12:40 gilleain at the moment, we are trying to replicate the old jcp functionality as faithfully as possible, before making major changes to how it works.
12:41 gilleain of, course, the nice thing about the design of the hub is that you can have 'hot swappable' functionality :)
12:47 CIA-16 cdk: jonalv * r13464 /cdk-qa/trunk/projects/081204-0001/ (3 files in 2 dirs): Added fingerprinter benchmark code
12:58 CIA-16 cdk: jonalv * r13465 /cdk/branches/jonalv-fingerprint-1.2.x/: yet another branch...
13:03 CIA-16 cdk: jonalv * r13466 /cdk/branches/jonalv-fingerprint/: removed variant based on trunk gonna use the one based on 1.2.x instead...
13:11 gilleain ugh:
13:11 gilleain if ((w | h | W | H) < 0) {
13:11 gilleain // At least one of the dimensions is negative...
13:11 gilleain return false;
13:11 gilleain }
13:12 gilleain curse you, bitwise operations!
13:28 jonalv egonw: Eclispe has it's own ant. I am yet to figure out how to get the latest ANt version into Eclipse
13:34 thomas_ku joined #cdk
13:34 egonw he thomas_ku!
13:34 thomas_ku moin
13:34 zarah hi thomas_ku
13:35 jonalv hm I will jsut run the build from terminal. Perhaps that works...
13:35 jonalv I wonder what it is doing...
13:35 jonalv it is taking quite a long time...
13:43 CIA-16 cdk: shk3 * r13467 /jchempaint/trunk/src/main/​org/openscience/jchempaint/ (4 files in 2 dirs): the applet is now detachable
13:46 egonw jonalv: yes: http://chem-bla-ics.blogspot.com/2​008/11/parallel-building-cdk.html
13:46 zarah egonw's link is also http://tinyurl.com/5qbqxj
13:47 CIA-16 cdk: shk3 * r13468 /jchempaint/trunk/src/main/org/opensc​ience/jchempaint/JExternalFrame.java: the applet is now detachable
13:47 jonalv egonw: uhm okey nice statistics :)
13:49 CIA-16 cdk: gilleain * r13469 / (3 files in 3 dirs): Calculates and resizes the screen better when the drawing goes over an edge.
13:52 CIA-16 cdk: goglepox * r13470 /cdk/branches/jchempaint-primary/src/main/org​/openscience/cdk/renderer/RendererModel.java: Changed getBondDistance() to getBondWidth() for RendererModel.getBondWidth()
13:53 egonw Gpox: ok, those two patches missed the sync
13:54 gilleain oops.
13:55 gilleain I kept all those methods to avoid having to do getRendererModel().getRenderi​ngParameters().getBondWidth()
13:56 gilleain but clearly, changing them made some bugs. :(
13:56 CIA-16 cdk: jonalv * r13471 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] first optimization brings down fingerprinting time to about 62.5% of what it was before
14:04 CIA-16 cdk: jonalv * r13472 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] first optimization brings down fingerprinting time to about 58.5% of what it was in the beginning
14:19 CIA-16 cdk: jonalv * r13473 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] third optimization brings down fingerprinting time to about 57.6% of what it was in the beginning
14:20 egonw mmm... I spot some pattern here...
14:22 jonalv egonw: you do? :)
14:22 * jonalv is firing up the profiler again now... :)
14:47 CIA-16 cdk: shk3 * r13474 /jchempaint/trunk/ (4 files in 3 dirs): more applet stuff
14:59 CIA-16 cdk: shk3 * r13475 /jchempaint/trunk/build.xml: cleaned up build.xml
15:09 CIA-16 cdk: gilleain * r13476 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/ (ISelection.java selection/ShapeSelection.java): Excluding bonds from the selection that are not inside the shape.
15:12 Julio joined #cdk
15:13 Julio left #cdk
15:45 CIA-16 cdk: jonalv * r13477 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] fourth optimization brings down fingerprinting time to about 55.1% of what it was in the beginning
15:45 CIA-16 cdk: egonw * r13478 /cdk/branches/jchempaint-primary/src/​main/org/openscience/cdk/controller/ (ControllerHub.java IChemModelRelay.java): Return what has been created
15:49 CIA-16 cdk: gilleain * r13479 /cdk/branches/jchempaint-primary/src/main/org/​openscience/cdk/controller/ControllerHub.java: Adding a new ring to the current moleculeset.
15:53 CIA-16 cdk: egonw * r13480 /cdk/branches/jchempaint-primary/src/main/org/o​penscience/cdk/controller/DrawBondModule.java: 1. Make use of powers of the new and improved IChemModelRelayAPI; 2. no excetion thrown, so no need to catch it
15:54 CIA-16 cdk: shk3 * r13481 /jchempaint/trunk/ (2 files in 2 dirs): looks like all the public applet functions are now working
16:00 edrin joined #cdk
16:16 CIA-16 cdk: jonalv * r13482 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] this optimization brings down fingerprinting time to about 53.3% of what it was in the beginning. (removed an extra call to StringBuffer.toString)
16:32 CIA-16 cdk: shk3 * r13483 /jchempaint/trunk/ (8 files in 2 dirs): created a test dir for the applets
16:51 edirn joined #cdk
17:03 edirn joined #cdk
17:04 carsten joined #cdk
17:05 CIA-16 cdk: jonalv * r13484 /cdk/branches/jonalv-fingerprint-1.2.x/src/main/or​g/openscience/cdk/fingerprint/Fingerprinter.java: [jonalv-fingerprint-1.2.x] fix for error in last commit
17:13 CIA-16 cdk: shk3 * r13485 /jchempaint/trunk/ (3 files in 2 dirs): load param works as well
18:34 jbrefort joined #cdk
18:38 slyrus_ joined #cdk
19:29 egonw joined #cdk
19:46 carsten joined #cdk

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