Camelia, the Perl 6 bug

IRC log for #cdk, 2008-12-01

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

All times shown according to UTC.

Time Nick Message
01:07 slyrus_ joined #cdk
01:31 tim_ joined #cdk
06:34 jbrefort joined #cdk
08:13 egonw joined #cdk
08:19 Gpox joined #cdk
08:52 CIA-16 cdk: egonw * r13398 /cdk/branches/cdk-1.2.x/build.xml: Added missing trim="true" attribute, fixing running test-all.
09:08 gilleain joined #cdk
09:16 jonalv joined #cdk
09:33 gilleain hi egonw, saw the jcp-primary nightly stuff, looks good. Didn't realise pmd was quite so fussy  about things like short variable names, but hey.
09:34 egonw :)
09:34 egonw seriously.... things add up...
09:34 gilleain true.
09:34 egonw when browsing a piece of code longer than 20 lines, short variable names often become a pain
09:34 egonw for example: "ac"
09:34 egonw atom container
09:34 egonw ok,
09:34 gilleain x, y
09:35 egonw of which point?
09:35 egonw short names often say what they are, but *not* the role
09:35 jonalv this reminds me of: http://www.linuxhq.com/kernel/v1​.3/53/Documentation/CodingStyle
09:35 zarah jonalv's link is also http://tinyurl.com/6zbmc2
09:35 egonw ac -> original container, or the one I am keeping as trash bin?
09:36 egonw x,y -> coords of an atom, or of the center of a ring...
09:36 egonw if they code is clean enough, it typically is clear...
09:36 egonw but in most CDK cases it is not
09:37 gilleain indeed.
09:37 gilleain did you look at the selection patch? or should I commit it (or throw it away?)
09:37 egonw yes, I did look at it...
09:37 egonw seems ok, but
09:37 egonw ...
09:38 egonw it only covers screen created selections, not?
09:38 gilleain yes
09:38 gilleain I wasn't sure what would be in common with programattic ones
09:38 gilleain er. programmatic
09:38 egonw the IChemModelRelay should likely have:
09:38 gilleain like "js> cdk.select("*.N")
09:39 egonw selectLasso(List<Point> lassoPoints)?
09:39 gilleain okay, sure, that's fine.
09:39 egonw selectRectangular(Point leftTop, Point rightBottom)?
09:39 egonw I guess we might have selectors like that...
09:39 gilleain I only put the logic in the controller to avoid changing the hub.
09:40 egonw but not sure yet if "*.N" should go into IChemModelRelay...
09:40 egonw was more thinking of the more general:
09:40 gilleain okay
09:40 egonw select(IAtomContainer)
09:40 gilleain where IAtomContainer is..
09:40 egonw well, I think something like selectLasso() in particular is good to have reusable...
09:40 gilleain another IAC?
09:41 egonw but I agree that that logic can perfercly well be placed in the controller and use select(IAtomContainer) too
09:41 egonw select(IAtomContainer) -> IAtomContainer would contain the selected atoms and bonds (and lone pairs, etc)
09:41 egonw well... so really it should be: select(IChemModel) I guess
09:42 egonw or not...
09:42 * egonw thinks...
09:42 gilleain hmmm.
09:42 egonw and is not sure
09:42 gilleain no me neither
09:43 egonw oh, also got a CDK Nightly running now, build with the opensource OpenJDK6:
09:44 egonw http://pele.farmbio.uu.se/nightly-1.2.x/
09:44 zarah egonw's link is also http://tinyurl.com/5nf6h5
09:44 gilleain ah good.
09:46 egonw getting some food, have not had breakfast yet
09:46 gilleain just eating mine :)
09:56 jonalv uhm cdk people: WHat do I do when I get "Could not create fingerprint: The AtomType Ge could not be found"?
09:56 thomasku joined #cdk
09:56 thomasku moin
09:56 zarah hi thomasku
09:57 gilleain hmmm. jonalv : but it gets all the other atom types right?
09:57 jonalv gilleain: seems so yea...
09:58 gilleain I mean Ge <i>is</i> an element
09:58 gilleain (oh well, html markup might have worked :)
09:58 jonalv yea I would think so too :)
09:58 jonalv that Ge is an element that is
09:59 gilleain I don't really know much about atomtyping, though.
09:59 gilleain I had a look at it the other day, but..
10:03 jonalv egonw: Can you help me?
10:05 gilleain he's gone for breakfast
10:06 jonalv oh aha
10:09 egonw jonalv: that means that the configuration found for Ge is not detected...
10:09 egonw now, the PubChem data is really dirty...
10:10 egonw so, you'll get lots of such errors...
10:10 jonalv are you telling me that it's not a cdk thing?
10:10 shk3 joined #cdk
10:12 gilleain so it's like Ge with 6 bonds or something weird?
10:13 egonw gilleain: right
10:13 egonw jonalv: could be... email me the PubChem CID
10:13 egonw and I'll check
10:13 egonw but many PubChem entries are not properly represented..
10:13 egonw and the SD files are even very bad
10:14 jonalv Okey you can's say anything from the error message?
10:14 egonw and many organometallic coordination bonds (like lone-pair-M interactions) are represented a SINGLE covalent bonds
10:14 jonalv There is a bit of computing involved in finding that error...
10:14 egonw CDK will not be able to understand the misuse of semantics there
10:14 jonalv I will do it again tomorrow I think...
10:14 egonw jonalv: therefore...
10:14 egonw you should catch the exception, not generate a fingerprint, and move on
10:15 jonalv egonw: right, all searching will be totally broken without a fingerprint though...
10:16 egonw well, the item will not show up, indeed...
10:16 egonw we can think about normalizing input later...
10:16 egonw that's the process of automatic data clean up before (or after) entering the db...
10:17 jonalv okey
10:17 * jonalv is profiling Brunn now :)
10:17 jonalv YourKit++
10:19 egonw gilleain: your patch looks fine
10:19 egonw for now at least...
10:19 gilleain okay
10:19 egonw we need to continue discussing what should stay on the controller side, and what needs to go into the relay hub
10:19 gilleain I've just been adding @module an copyrights
10:19 egonw bug please apply
10:20 egonw gilleain: ok, thanx :)
10:43 CIA-16 cdk: gilleain * r13399 / (13 files in 7 dirs): Selections.
10:43 gilleain ok, shk3 : it autoUpdates the selection now.
10:44 gilleain also, it's more jcp-like, in that it uses little squares.
10:44 egonw that's optional, I hope? I liked the filled circles
10:45 gilleain I did think about how to make it optional.
10:45 gilleain It's very easy to change.
10:46 shk3 adding @module to what?
10:46 gilleain what?
10:47 egonw cdk.module, right?
10:47 gilleain er. whoops.
10:47 gilleain mom
10:49 CIA-16 cdk: gilleain * r13400 /cdk/branches/jchempaint-primar​y/src/main/org/openscience/cdk/ (3 files in 3 dirs): Fixed module statements.
10:50 gilleain >_>, <_<
10:51 egonw oh, and the module names are 'control' and 'render'
10:51 egonw not 'controller' and 'renderer'
10:52 gilleain ack!
10:54 CIA-16 cdk: gilleain * r13401 /cdk/branches/jchempaint-primar​y/src/main/org/openscience/cdk/ (3 files in 3 dirs): Another fix to module statements.
10:54 gilleain right.
10:55 gilleain hmmm. there's a weird bug to do with the color of the selection elements.
10:56 gilleain when you move the mouse, they change from gray->black. hmmm.
10:57 gilleain Oh, and the thing with having filled circles is that it would cover the atom underneath.
10:57 gilleain unless the circle was semi-transparent, of course.
11:02 CIA-16 cdk: gilleain * r13402 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/ (6 files in 4 dirs): Fixed colors on selections.
11:34 jonalv lol: "Could not create fingerprint: The AtomType Ga.3plus could not be found"
11:34 jonalv are there no standards for this?
11:36 jonalv S.trioxide,
11:37 egonw "The AtomType Ga.3plus could not be found" that is not good
11:37 egonw like S.trioxide...
11:37 egonw those are CDK atom types...
11:37 egonw and recognized...
11:37 egonw but a look up fails...
11:37 jonalv oh
11:38 egonw please pastebin a stacktrace
11:38 egonw somewhere it tries to look up info on these atom types, but not using the correct atom type list there, I guess...
11:39 CIA-16 cdk: gilleain * r13403 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/ (3 files in 2 dirs): Added support for setting selection shapes as rectangles or circles.
11:40 jonalv egonw: http://pastebin.com/f2398d659
11:40 jonalv I am slurping your 1gb file. It seems to take 4 or 5 hours... :)
11:40 egonw that's too long...
11:40 egonw umm
11:41 egonw only those atom types?
11:41 jonalv egonw: but the memory use problem is fixed now
11:41 egonw no messages about C.sp3 or so?
11:41 egonw jonalv: excellent :)
11:41 jonalv egonw: no such message yet..
11:41 egonw ok...
11:42 egonw then it seems that the atom type list is outdated...
11:42 egonw compared to the CDKAtomTypeMatcher...
11:42 jonalv and I have done about 36000 molecules...
11:44 egonw no clue why you get these messages...
11:44 egonw but will update the atom type def list now
11:44 egonw and that should solve the pasted problems
11:44 jonalv oki
11:45 egonw I think something wrong with cdk-eclipse
11:46 jonalv egonw: okey...
11:47 egonw https://sourceforge.net/tracker2/?func=detail&am​p;aid=2370613&amp;group_id=20024&amp;atid=120024
11:47 CIA-16 cdk: gilleain * r13404 /cdk/branches/jchempaint-primary/src/main/org/​openscience/cdk/renderer/Renderer2DModel.java: Fixed infinite recursive loop I caused.
11:47 egonw gilleain++ (for r13404 :)
11:48 gilleain Serves me right for not doing it with a refactoring tool in the first place...
12:03 CIA-16 cdk: gilleain * r13405 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/g​enerators/BasicAtomGenerator.java: Compact mode rendering support.
12:04 gilleain should jcp options allow setting the compact status?
12:04 egonw guess so
12:06 shk3 doesn't do harm
12:06 shk3 on the other hand, in jcp, the compact mode is not of much importance, i would say
12:06 shk3 it's more for image generatino
12:06 gilleain true.
12:07 egonw in the application not
12:07 egonw but for the renderer most definately
12:07 egonw and not just for image generation
12:07 egonw SDF tables
12:07 egonw very useful to have compact mode then
12:07 gilleain true.
12:11 shk3 in the renderer, yes
12:11 shk3 that's what I wanted to say...
12:15 egonw I'm trying to focus... but my head just does not want to...
12:15 egonw I hate being ill :(
12:15 gilleain still ill? :(
12:15 jonalv :(
12:16 egonw yes, apparently so :(
12:16 gilleain all weekend then, sorry to hear that.
12:16 egonw during the weekend I felt better that now...
12:16 egonw it's lingering...
12:16 egonw not very pronounced presence...
12:16 egonw but just enough to not stay focused... and wondering off the the internet :(
12:17 egonw or do stupid things like tuning Nightly.... :(
12:18 gilleain tuning nightly is fairly worthwhile, I think.
12:18 egonw but not with deadlines on your mind
12:32 CIA-16 cdk: egonw * r13406 /cdk/branches/cdk-1.2.x/src/test/or​g/openscience/cdk/tools/DeduceBondO​rderTestFromExplicitHydrogens.java: Ummm... butaDIene has two double bonds, not?!?!?
12:32 CIA-16 cdk: egonw * r13407 /cdk/branches/cdk-1.2.x/src/main/org/​openscience/cdk/io/SMILESReader.java: Fixed 'accepting' of subclasses of classes that implement the supported IChemObjects.
12:33 CIA-16 cdk: egonw * r13408 /cdk/branches/cdk-1.2.x/src/test/or​g/openscience/cdk/tools/DeduceBondO​rderTestFromExplicitHydrogens.java: Removed unused import.
13:10 gilleain joined #cdk
13:35 carsten joined #cdk
14:12 CIA-16 cdk: egonw * r13409 /cdk/branches/cdk-1.2.x/src/main/org/open​science/cdk/layout/TemplateHandler.java: Fixed path to template list.
14:13 CIA-16 cdk: gilleain * r13410 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/v​isitor/TransformingDrawVisitor.java: Dashed wedges drawn properly.
14:23 jpansanel moin
14:23 zarah hi jpansanel
14:26 jonalv hi jpansanel
14:26 jpansanel hi jonalv
14:27 jonalv is there a E/R diagram for that database system somewhere?
14:27 jpansanel no
14:27 jpansanel I'll produce it tonight
14:27 jpansanel and give you a link tomorrow
14:27 jonalv oki :)
14:28 * jpansanel is working hard on the compilation of Mychem on Win32
14:28 jonalv okey what's the status on Mac?
14:36 CIA-16 cdk: gilleain * r13411 /cdk/branches/jchempaint-primary/src/main/org/​openscience/cdk/controller/AddBondModule.java: Removing stereo wasn't updating.
15:11 CIA-16 cdk: gilleain * r13412 /cdk/branches/jchempaint-primary/src​/main/org/openscience/cdk/renderer/ (4 files in 3 dirs): Aromatic rings now working.
15:26 jpansanel jonalv : I did not try on Mac
15:26 jonalv jpansanel: ah
15:34 jpansanel a user told me that It compiles successfully on Mac
15:34 jpansanel how ever, I've no further informations
15:34 jonalv good
15:34 jonalv I will try it out once I am ready for that :)
15:34 jpansanel cool
15:35 jpansanel If you can take some notes about the procedure
15:35 jpansanel I could update the documentation
15:36 jonalv jpansanel: yea well it won't be before christmus at least
15:36 jonalv I don't know when I will have time
15:36 CIA-16 cdk: gilleain * r13413 /cdk/branches/jchempaint-primary/src/​main/org/openscience/cdk/controller/ (ControllerModel.java ControllerParameters.java): Separating out controller parameters from the controller model.
15:45 jpansanel jonalv : ok
15:46 gilleain joined #cdk
15:47 jonalv egonw: ping
15:48 jonalv I wanna do a speed up bu switching from ArrayList to HashMap in FingerPrinter.java in cdk. What is the right approach?
15:56 rguha joined #cdk
15:57 gilleain hi rguha
15:57 rguha hi gilleain
15:58 rguha seeing lots of activity on the renderer :)
15:58 gilleain yup.
15:58 gilleain jchempaint is a good test case.
15:58 gilleain Or a good application, depending on your point-of-view :)
15:58 rguha :)
15:59 gilleain It's a little dispiriting to be re-creating old functionality that worked well, but nice to know that it is being done in a more extensible way (I hope).
16:00 rguha indeed
16:00 rguha btw, when you mention extensible, will it be configurable to the extent that a user can change width, spacing, colorof bonds, atoms etc?
16:00 gilleain in jcp?
16:01 rguha I was thinking that something like CSS for depictions would be very nice
16:01 rguha well for the renderer in general
16:01 rguha Rich Apodaca wrote about this idea sometime back
16:01 gilleain for the renderer in general, yes.
16:01 gilleain css for depictions. hmmm.
16:01 gilleain well, I do think that it wouldn't be so hard to generate svg server-side, and render it in the browser.
16:02 gilleain which would be nice.
16:02 rguha I was htinking that CSS would be more general than just browser based usages
16:02 gilleain css in a java application?
16:02 rguha yes
16:03 gilleain well, I suppose.
16:03 rguha but doesn't have to be CSS specifically - some form of textual markup controlling the depiction
16:03 gilleain hmmm.
16:03 rguha I just thought that repurposing CSS would be handy
16:03 rguha http://cssparser.sourceforge.net/
16:03 zarah rguha's link is also http://tinyurl.com/s4sh3
16:04 rguha just an idle idea :)
16:04 gilleain well, the intermediate layer is nearly a chemical-independant tree of objects, so DOM might work.
16:05 gilleain Sort of Group(Group(line1, line2, line3), Group("N", "C", "O"))
16:05 gilleain SVG was also suggested.
16:06 rguha nice
16:06 rguha S-expressions :)
16:07 egonw hi
16:07 zarah hi egonw
16:07 rguha hi egonw
16:08 jonalv egonw: is there a special procedure I should take before changing crusial parts of cdk code?
16:08 egonw yes, use a branch.
16:08 egonw that's all
16:08 jonalv I am gonna change 3 rows...
16:08 egonw make sure Nightly is happy about your code
16:08 egonw then send a patch
16:09 gilleain is it sort of:
16:09 egonw but making a clone on github should be fairly sheep (to follow the Tele2 joke :)
16:09 jonalv can I not run some test suite at my desktop?
16:09 gilleain -ArrayList listOfStuff;
16:09 gilleain + HashMap mapOfStuff;
16:09 gilleain ?
16:09 jonalv gilleain: exactly
16:09 jonalv that will speed up stuff alot...
16:10 gilleain cool (so long as it works :)
16:10 egonw ok, that should be fine, I guess
16:10 rguha egon, does build.xml work for you?
16:10 egonw yes: http://pele.farmbio.uu.se/nightly-1.2.x/
16:10 zarah egonw's link is also http://tinyurl.com/5nf6h5
16:10 jonalv I will try it out oncefreaking svn merge starts working
16:10 egonw rguha: using OpenJDK6
16:10 jonalv svn is borken. Jonas removed bin but svn removed src!
16:11 egonw hahahaha
16:11 jonalv _very_ scary!
16:11 egonw I think I've had that too in my early eclipse days )
16:11 egonw I think I've had that too in my early eclipse days :)
16:11 jonalv and he uses Ubuntu standard svn client
16:11 jonalv I just checked his terminal history he did remove bin
16:12 rguha egon, hmm not working here. Reports that I need ant 1.7, but I do have it
16:12 jonalv but bet.bioclipse.brunn.ui.src is gone!
16:13 egonw rguha: yes, weird
16:14 egonw rguha: where can I check the output of your 'ant -d build.xml' ?
16:14 egonw rguha: where can I check the output of your 'ant -f build.xml' ?
16:15 rguha one sec
16:20 rguha http://pastebin.com/m2d4e9ae1
16:20 egonw jonalv: you could ask rguha to run a benchmark of the array-FP version, and the map-FP version
16:20 rguha ant info output is at http://pastebin.com/d5b2cccc5
16:21 jonalv egonw: okey, I found this problem using YourKit++ btw :)
16:21 egonw jonalv: I think rguha originally convinced me to use it...
16:21 egonw rguha: was it not?
16:21 rguha yourkit rocks
16:21 egonw yes, agreed
16:22 egonw it plain does what you want it to do
16:22 jonalv I haven't tried the changed code yet though...
16:22 egonw WYWIWYG
16:22 egonw rguha: btw, I did not change Ant detection, so a bit puzzled...
16:23 egonw <!-- test Ant version -->
16:23 egonw <condition property="hasAnt17">
16:23 egonw <contains string="${ant.version}" substring="1.7"/>
16:23 egonw </condition>
16:23 egonw looks fine to me ...
16:24 rguha hmm, strange
16:24 egonw indeed
16:24 egonw is this using Nightly?
16:24 rguha can you run with JDK 1.5 and see if you get the same problem?
16:24 rguha no, just synced with SVN and ran it by hand
16:24 rguha nightly doesn't build either
16:24 rguha http://cheminfo.informatics.indiana.​edu/~rguha/code/java/nightly-1.2.x/
16:24 zarah rguha's link is also http://tinyurl.com/6xnh2n
16:29 egonw rguha: works here :(
16:30 rguha aargh :(
16:33 egonw git bisect?
16:33 rguha not using git yet :-/
16:34 egonw rguha: I'm going to shutdown anynomous bug reporting for the CDK
16:34 egonw third spam in the bug tracker now...
16:34 egonw now on an open bug :(
16:35 rguha good idea
16:37 rguha what version of ant-contrib do you have?
16:37 egonw the one you uploaded
16:37 egonw did not change version of that
16:38 rguha hmm
16:38 rguha doesn't compile on OS X either, same error
16:39 rguha after commenting out the hasAnt17 test, I get the following error: /Users/rguha/src/java/cdk/bran​ches/cdk-1.2.x/build.xml:343: Excludesfile /Users/rguha/src/java/cdk/branches​/cdk-1.2.x/build/ant1.6.javafiles not found.
16:44 egonw mmm...
16:44 egonw not that again...
16:44 egonw that files does not exist
16:44 egonw but it should not complain about it...
16:44 egonw what build.xml revision do you have?
16:45 egonw 13398?
16:45 rguha 13413
16:45 rguha it cmplained because I removed the hasAnt17 check to see why it failed
16:45 rguha the weird thing is that <fail unless=ahsAnt17 ... /> fails
16:45 rguha but hasAnt17 is true
16:45 egonw ah... yes, that explains...
16:46 egonw (in reply to <rguha> it cmplained because I removed the hasAnt17 check to see why it failed)
16:46 egonw yes, that's really weird indeed...
16:46 rguha anf the error is repoted after the foreach fails
16:46 rguha I have ot go now, will look at it next week
16:47 egonw I'm running ant 1.7.1
16:47 egonw maybe 1.7.0 has a foreach related bug?
16:47 rguha hmm, I'll try that
16:51 egonw I don't have access to 1.7.0 here... and cannot easily verify that it won't work with 1.7.0 go me either
16:51 rguha ok, it's compiling with 1.7.1
16:52 egonw ok, then I'll bump version checking to not acccept 1.7.0
16:52 rguha about to commit the fix
16:52 egonw ok, thanx
16:53 rguha done
16:53 CIA-16 cdk: rajarshi * r13414 /cdk/branches/cdk-1.2.x/build.xml: Updated build script to indicate that ant 1.7.1 or better is required
16:55 rguha bye
16:59 CIA-16 cdk: shk3 * r13415 /jchempaint/trunk/ (7 files in 2 dirs): the applet build (just) works
17:35 jbrefort joined #cdk
18:30 slyrus_ joined #cdk
18:52 carsten joined #cdk
19:21 CIA-16 cdk: gilleain * r13416 /cdk/branches/jchempaint-primary/src/​main/org/openscience/cdk/controller/ (ControllerModel.java ControllerParameters.java): Added cdk.module statement.
19:32 edrin joined #cdk

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