Camelia, the Perl 6 bug

IRC log for #cdk, 2008-04-30

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

All times shown according to UTC.

Time Nick Message
04:19 jbrefort joined #cdk
05:41 thomas_ku joined #cdk
05:42 thomas_ku moin
05:46 egonw joined #cdk
05:53 rojasm joined #cdk
06:10 egonw_ joined #cdk
06:17 egonw__ joined #cdk
08:03 rojasm egonw how should I show the comparison errors. Doing ant test-all for each one and then copy paste?
08:04 egonw yes
08:05 egonw http://chem-bla-ics.blogspot.com/2007/11/​comparing-junit-test-results-between.html
08:07 rojasm that is fantastic report ;)
08:07 rojasm thanks
08:08 rojasm sourceforge is down, isn't it? . I cannot do any update
08:09 egonw was working earlier this morning for metware
08:09 egonw mom
08:12 egonw well, the website seems down
08:14 egonw ummm... slashdot does not work either
08:16 egonw this likely means we lost contact with the US?
08:31 jpansanel moin
08:31 * egonw just loves submitting NMR spectra to nmrshiftdb.org :)
08:32 egonw so relaxing
08:32 egonw hi jpansanel
08:32 * egonw will hit the 900 submits today! all manual !
08:43 egonw I mean, it's so rewarding to see a computer learn because you teach it :)
08:44 egonw we don't need no stinkin' students...
08:44 egonw hahahaha
08:44 egonw (never mind :)
09:50 egonw slashdot is still not working for me
09:50 rojasm egonw: do you think it could be more effective if the IP caculation has a own module
09:50 egonw well, easier to see bug stats, missing unit tests etc...
09:54 rojasm because one of the ideas that you say was that descriptor classes should not contain the algorithm self
09:54 egonw if you do that, I'd suggest modules ip and qsar-ip
09:54 egonw right, so 'ip' with the algorithms
09:54 egonw and qsar-ip with the wrapping into a QSAR descriptor
09:55 rojasm I am think the IP needs first to recognize functional groups
09:55 egonw for which you can use SMARTS
09:56 rojasm ok
09:57 jbrefort joined #cdk
09:57 rojasm and the process for the decision?
09:57 rojasm using still the weka tree
09:58 egonw if converted to Java code, yes, that would be fine for now
09:58 rojasm ok
09:59 rojasm it doesn't work still the cdk.sourceforge.net, :(
09:59 egonw same here
10:02 rojasm other  question about the HOSE code in cdk
10:03 rojasm is it only usable for atoms or you can put what you want as Atomtype
10:03 egonw don't know
10:05 * egonw has to do some household stuff now
10:05 egonw bbl
12:36 egonw_ joined #cdk
12:36 * egonw_ expects a major dns server is malfunctioning
12:58 egonw_ rojasm: sf.net seems to be available again
13:22 egonw joined #cdk
13:23 egonw joined #cdk
13:28 rojasm yes thanks
13:28 rojasm Know I continue
13:30 CIA-30 cdk: miguelrojasch * r10725 /cdk/branches/miguelrojasch-charges/src/​test/org/openscience/cdk/graph/invariant​/ConjugatedPiSystemsDetectorTest.java: correction in tests which were failing because they could not be read
13:30 egonw :)
13:32 CIA-30 cdk: miguelrojasch * r10726 /cdk/branches/miguelrojasch-charges/src/ (4 files in 2 dirs): changing the results of some tests. Added Inchi ID. Substituted creating the object with the builder than with smiles parser
13:33 rojasm but the transmission is incredible slow
13:33 rojasm mmm man kann nicht alles haben
13:34 egonw yeah, it indeed has been slow lately..
13:34 egonw btw: subversive seems to be the best SVN plugin for eclipse now:
13:34 egonw http://wiki.eclipse.org/Subversive_Plan
13:49 CIA-30 cdk: miguelrojasch * r10727 /cdk/branches/miguelrojasch-CMLReact/src/main/o​rg/openscience/cdk/qsar/descriptors/molecular/ (4 files): Merged the branch.miguelrojasch-CMLReact from cdk.trunk, rev: 10726
13:52 CIA-30 cdk: miguelrojasch * r10728 /cdk/branches/miguelrojasch-charges/src/main/o​rg/openscience/cdk/qsar/descriptors/molecular/ (4 files): Merged the branch.miguelrojasch-charges from cdk.trunk, rev: 10726
14:14 CIA-30 cdk: egonw * r10729 /cdk/trunk/src/ (7 files in 4 dirs): Moved structgen atom type code to the structgen module
14:16 CIA-30 cdk: egonw * r10730 /cdk/trunk/src/main/org/openscience/cdk/​atomtype/StructGenAtomTypeGuesser.java: Fixed copyright years (cheap commit to test why Revision: updates are not working...)
14:25 CIA-30 cdk: miguelrojasch * r10731 /cdk/branches/ (2 files in 2 dirs): miss in the ReactionSchemeTest initiating the chemObject
14:27 CIA-30 cdk: egonw * r10732 /cdk/trunk/src/main/org/openscience/cdk/​atomtype/StructGenAtomTypeGuesser.java: Do I really need to use svn:keywords ??
14:27 jbrefort joined #cdk
14:33 CIA-30 cdk: egonw * r10733 /cdk/trunk/src/main/org/openscience/cdk/​atomtype/StructGenAtomTypeGuesser.java: Is this a client setting?? (HOME/.subversion/config)
14:34 egonw no. it is not either... :(
14:34 egonw bbl
14:53 rojasm egonw I see a loot of smarts which fails?
14:53 egonw yeah, about 36
14:53 rojasm do you think it is a good Idea to take that for the IP
14:53 egonw http://cheminfo.informatics.indiana.edu/~rgu​ha/code/java/nightly/test/result-smarts.html
14:54 egonw I guess it depends on which SMARTS features you need...
14:54 egonw ask on cdk-devel what features are currently not supported, causing failing tests...
15:03 CIA-30 cdk: egonw * r10734 /cdk/trunk/src/ (9 files in 4 dirs): Moved two other atom type matchers (unstable) to extra, to get the atom type module stable
15:09 rojasm do you know in what jar is contained the module isomorphism. if not I look for not problem
15:10 rojasm nothing it is self a jar
15:10 rojasm :)
15:11 egonw when looking in which jar a certain class is, I always do this:
15:11 egonw grep ClassILookFor build/*.javafiles
15:11 rojasm because the GasteigerPEPEPartialcharges needs the StructureResonanceGen. and this needs isomorphism
15:12 rojasm that means each qs...jar needs the dependency with isomorphism.jar or only the cdk-charges
15:14 rojasm mmm
15:15 egonw on both
15:15 egonw the .cdkdepends basically define a classpath
15:22 CIA-30 cdk: egonw * r10735 /cdk/trunk/src/main/org/openscience​/cdk/config/data/cdk_atomtypes.xml: Removed duplicate entry
15:48 egonw rojasm: when comparing unit test results, make sure to compare against trunk version 10726
15:48 egonw the one you synched against...
15:51 rojasm yes
15:53 CIA-30 cdk: miguelrojasch * r10736 /cdk/branches/ (2 files in 2 dirs): Removed one parameter which produced error in compilation
15:53 CIA-30 cdk: egonw * r10737 /cdk/trunk/src/test/org/openscience/cdk/ (CoverageAnnotationTest.java CoverageTest.java): Take advantage of src/main, src/test split up: protected methods should now be tested too
15:54 * egonw is away for making/eating dinner
17:23 CIA-30 cdk: miguelrojasch * r10738 /cdk/branches/ (6 files in 2 dirs): added dependencies in qsar jars
17:31 egonw joined #cdk
17:37 CIA-30 cdk: miguelrojasch * r10739 /cdk/branches/ (8 files in 4 dirs): changed extends for the corresponding test qsar
17:45 egonw bbl
18:00 jonalv joined #cdk
19:12 egonw joined #cdk
19:12 rojasm egonw: around
19:12 egonw yes
19:12 rojasm I have a big confusion
19:12 egonw just worked a bit on MetWare...
19:12 egonw using a REST set up :)
19:13 rojasm with CDKTestCase and ATomicDescriptorTest
19:13 egonw Jim (Camb) had a nice remark on my last blog item...
19:14 rojasm should all descriptor with only test-qsaratomic model be AtomicDescriptorTest or
19:14 rojasm those which are in de qsar.atomic packages
19:15 egonw not sure what you mean...
19:15 rojasm or better question why some atomic descriptors are extends of CDKTestCase and others from AtomicDescriptorTest
19:16 rojasm should all of them be AtomicDescriptors
19:16 egonw ah, they might not all have been transformed yet?
19:16 egonw or, some atomic descriptors are not truely atomic yet?
19:16 egonw got an example?
19:17 rojasm I remember we had this discussion
19:17 rojasm between atomic descriptor
19:17 rojasm which means calculation for the atom independent of the container
19:17 rojasm for example covalentreadius
19:18 rojasm or atomcalculation depending of the atomcontainer for example SigmaElectronegativity
19:19 rojasm but the question is: all descriptors which are contained in qsar.atomic package should be in qsaratomic module and extends the test from IATomicDescriptor, isn't it?
19:20 rojasm and the same for bonds?
19:20 egonw mom...
19:21 rojasm no problem
19:24 egonw ok, back now...
19:24 egonw yes, and indeed the same for bonds, but then IBondDescripto
19:25 rojasm perfect
19:25 egonw you might notice that some atom and bond descriptors are currently in the qsar-molecular module...
19:25 jbrefort joined #cdk
19:25 egonw that's because they depend on atom or bond descriptors...
19:25 rojasm yes
19:25 egonw but that would be solved by separating algoritms from descriptor wrapping...
19:26 rojasm ok. they are as molecular module because have dependences
19:26 rojasm ok I look at
19:27 rojasm doubts solved
19:29 rojasm and why the others atomic descriptors if they depend only from the extension AtomicDescriptorTest neveertheless they have a test with Test suite method?
19:29 rojasm should they have it?
19:30 egonw example?
19:31 rojasm public static test suite(){blabla} for AtomDegreeDescriptorTest for example
19:32 egonw ok, mom
19:33 rojasm know I see in DescriptorTest the explication for that
19:33 egonw you mean why AtomDegreeDescriptorTest  has a suite() method?
19:33 rojasm yes
19:33 egonw yes, the setDescriptor() call is important
19:34 egonw http://news.slashdot.org/n​ews/08/04/30/1455205.shtml
19:34 egonw SF and Slashdot share a router...
19:35 egonw and both were not accessible for some hours :)
19:35 egonw so, we noted :)
19:35 rojasm jeje
19:35 egonw the message also notes that they are moving to better/faster hardware..
19:35 egonw I hope that applies to SF too, and not just to Slashdot
19:36 rojasm mmm
19:38 * egonw will try to get rcdk SVN checked out
19:40 rojasm sorry other question. with the qsar.bond MassNumberDifferenceDescriptorTest should be have the suite() method, isn't it?
19:42 egonw mmm... seems it is not being included in MqsarbondTests
19:42 egonw ah, no
19:42 egonw see cdk.modelsuites.MqsarbondTests
19:42 egonw it's a JUnit4 class
19:43 egonw no, it is not ...
19:43 egonw huh...??
19:44 egonw anyway...
19:44 egonw all 9 tests are run here...
19:44 egonw both from the suite as well as directly from the class itself...
19:46 rojasm should I include it one
19:46 egonw no
19:46 rojasm or that means that is not necessary for the rest either
19:46 egonw all tests run, so it seems to do fine without the suite method...
19:47 rojasm and if you make the same for atomic qsar?
19:47 egonw well... let's split algorithms from descriptor wrappers first...
19:47 rojasm ok
19:58 CIA-30 cdk: miguelrojasch * r10740 /cdk/branches/miguelrojasch-charges/src/t​est/org/openscience/cdk/qsar/descriptors/ (9 files in 2 dirs): added corresponding extends for qsar atomic and bond descriptor
20:00 rojasm egown  ; is it better if first I merge from my brunches.miguelrojasch-CMLReaction to branches.miguelrojasch-charges?
20:02 egonw we tried that last time.. might be tricky...
20:02 egonw but the changes might not be as large now...
20:03 egonw I'd rather see both patches disected, and then merged with trunk/
20:05 rojasm disected?
20:05 egonw analyzed
20:05 rojasm you mean one branch to other and the to trunk
20:05 rojasm and then the trunk
20:06 egonw I mean, merge CMLReaction to trunk and charges to trunk separately...
20:07 rojasm :)
21:01 * egonw goes to bed
21:01 egonw by rojasm
22:20 CIA-30 cdk: miguelrojasch * r10741 /cdk/branches/miguelrojasch-charges/src/main/org/​openscience/cdk/qsar/AbstractBondDescriptor.java: added isCachedAtomContainer method

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