Camelia, the Perl 6 bug

IRC log for #cdk, 2010-06-17

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

All times shown according to UTC.

Time Nick Message
05:12 egonw joined #cdk
05:29 egonw_ joined #cdk
06:15 egonw_ joined #cdk
06:37 carsten joined #cdk
06:40 jbrefort joined #cdk
06:46 Gpox joined #cdk
07:18 sneumann_ joined #cdk
11:49 maclean joined #cdk
11:49 maclean hi
11:49 zarah saluton maclean
11:53 egonw_ hi maclean
11:54 maclean hi - just finished re-reading "git from the bottom up". Probably understood it this time...
11:55 egonw_ :)
11:55 egonw_ cool
11:57 maclean git show-branch is particularly eye-opening, as it shows some of the horrible things I've been doing to my cdk branches.
11:59 egonw_ :)
11:59 egonw_ that command is new to me
12:00 egonw_ btw... will merge in your patch, and rerun nightly shortly...
12:00 egonw_ but need to finish a blog post first
12:01 maclean ok
12:15 maclean nice screenshots.
12:15 sneumann_ Does anyone know a commandline tool to calculate exact mass from an elemental composition ?
12:16 sneumann_ I tried
12:16 sneumann_ echo "InChI=1/C11H16N2/" | babel -in - -out sdf |  obprop -
12:16 sneumann_ but obviously that's not a proper inchi
12:16 sneumann_ to make an SDF :-(
12:16 sneumann_ I think CDK needs a set of commandline tools
12:16 sneumann_ possibly autocreated ?
12:16 maclean There are cdk command line tools
12:17 maclean in a project with a horrible name.
12:17 maclean mom
12:18 maclean "clapps" :)
12:18 sneumann_ thanks
12:18 maclean http://cdk.svn.sourceforge.​net/viewvc/cdk/cdk-clapps/
12:18 zarah maclean's link is also http://tinyurl.com/3x9ohq5
12:19 maclean hmmm. no formula calculator though.
12:19 sneumann_ :-(
12:19 sneumann_ Ah, I'll try to think about beanshell...
12:19 egonw_ maclean: yeah, nice JChemPaint uses :)
12:20 egonw_ sneumann_: ah, email to the cdk-user@ mailing list...
12:20 egonw_ Vincent recently took over the command line tools, to give it renewed effort
12:21 egonw_ he will very much appreciate to hear from the community that such work would be used
12:21 egonw_ maclean: clapp is something nasty, not?
12:21 egonw_ well, not according to google
12:21 maclean "The clap" is slang for a sexually transmitted disease
12:22 egonw_ right
12:22 egonw_ but with one p
12:22 maclean Well, ok, yes :)
12:22 maclean http://en.wikipedia.org/wiki/Gonorrhea
12:22 zarah maclean's link is also http://tinyurl.com/nacm7
12:22 egonw_ :)
12:23 maclean 16th century cure : mercury injections...
12:24 egonw_ :)
12:25 egonw_ ah, well... makes you wonder how we think about current medicine in 100 years :)
12:25 egonw_ http://en.wikipedia.org/wiki/Female_hysteria
12:25 zarah egonw_'s link is also http://tinyurl.com/ffn4q
12:25 egonw_ the cure here was a... vibrator
12:25 egonw_ now we have viagra
12:25 egonw_ weird cures are of all time
12:25 egonw_ well, the best of all, I guess:
12:26 egonw_ homeopathy
12:27 maclean At least homeopathy did you no harm (or good). Much better than bloodletting.
12:28 maclean Anyway. What do you think about including the source jar for signatures in the cdk?
12:28 egonw well, I can tell you funny stories about homeopathy being harmless or not...
12:28 maclean It would probably have to be a separately reviewed commit as it would have to touch javadoc.xml
12:29 egonw it is not
12:29 egonw and the story is not really funny either
12:29 egonw why does it need to touch javadoc.xml?
12:29 egonw as to include the jar?
12:30 * egonw thinks he should add the deps to the classpath when calling <javadoc>
12:30 maclean first to unpack the jar to a temp dir, then to add the unpacked dir to sourcepath, then to delete the temp dir
12:31 maclean unless there is a way to give the javadoc tool a jarfile
12:35 maclean another minor thing is that the javdocs are not generated with reference to the java sources
12:36 maclean so that there will be "java.lang.Integer" everywhere instead of just "Integer"
12:36 egonw mom, let me check
12:36 egonw ah, just saw a solution for that:
12:36 egonw http://firstclassthoughts.co.uk/ant/javadoc.html
12:36 zarah egonw's link is also http://tinyurl.com/3ywcfwl
12:36 egonw last paragraph :)
12:38 egonw I think we can just use <classpath>:
12:38 egonw http://ant.apache.org/manual/using.html#path
12:38 zarah egonw's link is also http://tinyurl.com/5ew3gs
12:40 maclean hmmm. I don't think so. The <javadoc> tag only supports nested <sourcepath> tags (or <source> or <fileset>)
12:40 maclean I tried using zipfileset or somesuch, but that didn't work.
12:40 maclean I'll try classpath, though..
12:40 egonw http://ant.apache.org/manual/Tasks/javadoc.html
12:40 zarah egonw's link is also http://tinyurl.com/24c73g8
12:41 maclean Yeah, I did look at the ant manual when trying to work this :)
12:42 egonw :) that was not implied!
12:42 maclean I wasn't correct when I said that it doesn't support classpath - there's one there already
12:43 maclean but it doesn't load the sourcefiles that way.
12:46 egonw perhaps you can ask on stackoverflow?
12:47 maclean Could do. It does work if just unpacked to a tmp dir, I would just prefer a neater solution!
12:48 egonw indeed, that's you can ask for
12:48 * egonw is applying maclean's patches
12:48 maclean cool
12:49 maclean ooh. that <link> trick works nicely.
12:49 egonw cool
12:49 egonw can you make that a separate patch?
12:49 egonw with just that?
12:50 egonw than I can apply it to cdk-1.2.x and mastet
12:50 egonw than I can apply it to cdk-1.2.x and master
12:50 egonw maclean++ (for testing!)
12:50 maclean what, just the one-line link thing?
12:50 * sneumann_ now has a working bsh snippet
12:50 egonw yeah, just the adding of the <link> to javadoc.xml
12:51 egonw sneumann_++
12:51 maclean ok
12:51 egonw sneumann_: try groovy next time
12:52 maclean cinfony is another possibility
12:52 maclean http://code.google.com/p/cinfony/
12:52 zarah maclean's link is also http://tinyurl.com/melth7
12:52 egonw maclean: cinfony does not allow dhacking, or?
12:52 maclean dhacking?
12:52 egonw umm... -
12:52 egonw umm... -d
12:53 maclean I thought it allowed scripting.
12:53 maclean It's python bindings to various projects, including foreign non-CDK ones. (which are of course forbidden!)
12:54 egonw on this channel? sure is ;)
12:55 maclean I found it a little irritating that one reply to cdk-discussion was basically "the solution to your problem is to use our toolkit instead"
12:55 maclean On the other hand, it did turn out to be quite good code (except in c).
12:56 egonw :)
13:02 maclean Argh. No. That's not what I wanted.
13:03 maclean The signature package (signature.DAG, signature.AbstractVertexSignature, etc) are in the javadocs for cdk.
13:03 maclean Seems obvious now I say it but still.
13:04 maclean There may not even be a way to inherit javadocs without doing this :( :(
13:09 maclean Ahhh...."Contrary to the documentation, you can’t produce Javadoc on source in jar files. "
13:10 maclean "Similarly you can’t produce Javadoc from jars of class files, not even rudimentary Javadoc. "
13:10 egonw ah, that sucks
13:11 egonw Nightly is rerunning
13:18 maclean Ok
13:18 maclean gjdoc looks interesting
13:18 maclean Sample output: http://developer.classpath.org/doc/
13:18 zarah maclean's link is also http://tinyurl.com/2wtyrvq
14:05 mgerlich joined #cdk
14:48 carsten joined #cdk
15:23 maclean joined #cdk
15:37 jonalv joined #cdk
15:37 jonalv I am looking at org.openscience.cdk.io.FormatFactory.guessFormat
15:38 jonalv if it fails to determine the format, how will it signal that?
15:40 maclean hmmm
15:43 maclean It returns null, from what I can see.
15:43 jonalv yea adding that to javadoc now
16:13 jpansanel joined #cdk
16:32 maclean wow. http://raphaeljs.com/
17:03 egonw @tell maclean why do I have the feeling you are going to write a RaphaelRenderer?
17:03 zarah Consider it noted.
17:17 sneumann_ joined #cdk
17:33 maclean joined #cdk
17:35 CIA-47 cdk: maclean cdk-1.2.x * r1f8bb2d / javadoc.xml :
17:35 CIA-47 cdk: Improved javadoc generation using a link tag, so that references to java library classes are resolved properly
17:35 CIA-47 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://bit.ly/a415iG
17:35 maclean egonw : all those spelling errors are false positives
17:35 zarah maclean: You have new messages. Write '/msg zarah @messages' to read them.
17:35 egonw maclean: indeed
17:35 egonw so, no worries about those
17:36 egonw those are not blockers...
17:36 egonw I'm working on a patch for OpenJavaDocCheck to remove false positives...
17:36 maclean Tricky.
17:36 egonw it should:
17:36 egonw 1. not analyze tags
17:36 maclean One easy one should be the "gt"
17:36 egonw 2. not analyze HTML elements
17:36 egonw yes, like that gt
17:36 egonw 2. not analyze HTML elements and entities
17:36 egonw likely from: &gt;
17:37 maclean yeah, it was.
17:37 egonw it should also not test stuff in <code></code> etc
17:37 egonw 3. not fail on class names
17:37 egonw but I am not sure yet on how to do all that
17:37 egonw easier is names of chemicals :)
17:37 egonw that is just another dictionary
17:38 egonw were there true positives too?
17:38 maclean Does it break up method names by camel case?
17:38 egonw dunno...
17:38 egonw the spell checker works on what JavaDoc returns...
17:39 maclean No, no true positives that I can see.
17:39 egonw need to look into what it returns, and if I can use others formats
17:39 egonw maclean: so, only those PMD warnings for the tests left...
17:39 egonw I am about to look at those
17:39 egonw just applied your javadoc.xml patch to cdk-1.2.x
17:39 maclean Saw that, yes.
17:39 maclean The PMD warnings are mostly false.
17:40 maclean Except MoleculeFromSignatureBuilderTest
17:40 maclean which I don't understand.
17:40 maclean Since it does have unit tests.
17:40 egonw let me check
17:52 egonw sorry...
17:52 egonw my WS needed refreshing...
17:53 egonw ok, the first is indeed a false positive of PMD
17:54 egonw second is a TP, fix pending
17:54 egonw public void testCoverage() {
17:54 egonw Assert.assertTrue(super.runCoverageTest());
17:54 egonw }
17:55 egonw third is also a TP
17:55 egonw in AtomSignatureTest
17:55 egonw poh, that reminds me... please remove all output to STDOUT
17:55 egonw poh -> oh
17:55 egonw see:
17:55 egonw http://pele.farmbio.uu.se/nightly-si​gnatures/test/result-signature.html
17:55 zarah egonw's link is also http://tinyurl.com/382rmyl
17:56 egonw Standard Output section should be empty
17:57 egonw maclean: btw, I think the patch can be submitted for review tonite
17:57 egonw think is understatement
17:57 maclean ok, good.
17:59 maclean None of the rest of the CDK follows these rules - full unit test coverage, full javadocs, no test stdoutput, ...
17:59 egonw define rest
17:59 egonw the core does
17:59 maclean Pfft. Maybe.
17:59 maclean :)
17:59 egonw but true, the fast majority does not YET
17:59 egonw but that is really the goal
17:59 egonw and I have worked hard towards that goal
18:00 egonw but there is so much code
18:00 maclean There is.
18:00 egonw so, the least we can do is ensure all new code does not make the situation worse
18:00 maclean Well true.
18:05 egonw anyway... all these stability rules are what is the best we could come up with...
18:05 egonw it is not ideal... if you have ideas how some of those rules can be replaced by better ones to aim at the same goal, that is much appreciated!
18:07 maclean No, they're good rules, I just worry that excessive bureaucracy will put people off contributing.
18:07 egonw agreed
18:08 egonw on the other hand, these rules show that we take quality serious, which is appreciated by others...
18:08 egonw but this is why I help people with fixing all those issues
18:08 maclean One thing I realised is that the tests in the cdk are a mix of unit tests and functionality tests
18:08 egonw and all senior CDK developers are very much invited to do the same
18:08 egonw and provide new developers with patches that show how the code can be improved
18:08 egonw that's the idea anyway
18:09 egonw maclean: yes, they indeed are
18:10 egonw https://sourceforge.net/tracker/?func=detail&amp​;aid=3017699&amp;group_id=56262&amp;atid=479921
18:10 egonw bug report with PMD
18:11 egonw the fourth PMD warning is a false positive too, but cannot figure out why that fails
18:13 egonw btw, what about MoleculeSignatureTest lines 112-121
18:13 egonw that seems some left over?
18:13 maclean mom
18:14 egonw the MoleculeSignatureTest warning seems a TP
18:14 egonw you can set a time out for that test
18:14 egonw to not exceed 5 seconds or so...
18:14 maclean I'll just remove it.
18:15 egonw that's an option too :)
18:15 egonw make sure it was not listed in any @TestMethod :)
18:15 maclean It wouldn't have been.
18:16 egonw the problems in SignatureQuotientGraphTest are actually TPs too...
18:16 egonw the problem is actually in line 25 :)
18:16 egonw maclean: no, guess not
18:17 maclean oh, wrong import?
18:19 maclean Hmmm. Doesn't look like it.
18:19 egonw it's the JUnit3 Assert, versus the JUnit4 Assert which is in the org.junit package :)
18:21 maclean Fixed.
18:21 maclean Er, but not commited or pushed.
18:24 egonw want to sent it this way?
18:25 maclean Well, I'm also removing sys.out and fixing those other PMD warnings.
18:26 egonw ok, let me know when I can pull them in
18:26 jpansanel joined #cdk
18:35 CIA-47 cdk: maclean master * r1f8bb2d / javadoc.xml :
18:35 CIA-47 cdk: Improved javadoc generation using a link tag, so that references to java library classes are resolved properly
18:35 CIA-47 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://bit.ly/a415iG
18:35 CIA-47 cdk: Egon Willighagen master * r055f702 / (6 files in 3 dirs): Merge branch 'cdk-1.2.x' - http://bit.ly/cpq9gp
18:35 CIA-47 cdk: Egon Willighagen master * r9d679fe / javadoc.xml : Removed two lines added in master to the removed doccheck target - http://bit.ly/d6YTEz
18:46 maclean Is there a way to recompile just one module, so that ant -Dmodule=X test-module works?
18:47 egonw yes
18:47 egonw ant -Dmodule=X compile-module
18:48 maclean Hmmm. Ok. It's not in ant -projecthelp
18:48 maclean ...because the task does not have a description.
18:49 egonw sounds like a patch
19:15 sneumann__ joined #cdk
19:33 maclean Nope, that compile-module is no good for me. Unless there is some other target I'm missing...
19:34 maclean I want to alter the tests, then recompile the module, then check the results. Not re-run test-all after clean.
19:34 egonw ah, use this:
19:34 egonw ant -Dmodule=test-X compile-module
19:34 maclean ah...
19:35 maclean anyway, all stdout gone.
19:36 egonw got a commit for you:
19:36 maclean Where do the PMD results go?
19:36 egonw reports/pmd/
19:36 egonw here's the commit:
19:36 CIA-47 cdk: Egon Willighagen master * rc94ebcf / pmd/customCDK.xml : Added PMD tests for detecting misuse of TestClass and TestMethod, e.g. TestClass on a method. (fixes #3014808) - http://bit.ly/a6vitd
19:36 maclean Ah, excellent!
19:39 maclean ant -Dmodule=signature qa-module ?
19:39 maclean (for PMD)
19:40 egonw or use this:
19:40 egonw ant -f pmd.xml -Dpmd.test=custom -Dmodule=X test-module
19:42 maclean Hmm. Either of those give me empty reports/pmd/signature.xml files.
19:43 egonw then there are no errors left, I think
19:43 maclean Ah, good.
19:46 maclean Fixes to tests pushed.
19:47 egonw one commit ,correct?
19:47 maclean One test had to be commented out (sadly, the integer invariants) as it can't be tested for yet.
19:47 maclean One commit, yes.
19:48 egonw I *love* btw that I do not even need to fetch from your github repos to get that one patch:
19:48 egonw wget http://github.com/gilleain/cdk/commit/2a95​a87395a38a05ee7078fbf382f105ab683478.patch
19:48 zarah egonw's link is also http://tinyurl.com/36k6k6n
19:48 egonw git am -3 2a95a87395a38a05ee7078fbf382f105ab683478.patch
19:48 egonw and done :)
19:49 maclean Hmmm. Interesting.
19:49 egonw ok, running: ant clean dist-all test-dist-all
19:49 maclean When are you meeting Lars tomorrow?
19:49 egonw and if no errors, then I will push and rerun Nightly
19:50 egonw ummm... 10:30~ish or so
19:51 maclean Ok, just wondering if it made sense for me to try and implement some chiral stuff before then, but probably not.
19:52 maclean I think that TetrahedralChiral to [C@@] can largely be done, although the other way round will be a bit hacky.
19:52 egonw it was for SMILES too
19:52 egonw the problem is that when you encounter the @@, you do not know the involved atoms yet
19:53 maclean Ah, yes, I remember reading your solution in the smiles patches.
19:53 maclean Actually, I hadn't even considered that problem :)
19:55 * Nightly is working on the signatures branch
19:57 bag_ joined #cdk
20:00 maclean IStereoElement has no methods? :(
20:01 egonw correct
20:01 egonw reason: there is nothing in common
20:01 egonw E/Z, tetrahedral have just different things
20:02 egonw so, you need to check the subinterface
20:02 maclean oh ITetrahedralChirality.
20:02 egonw right
20:03 maclean That's close. I almost added a depedency on the data module with TetrahedralChirality
20:04 maclean :)
20:04 egonw no worries
20:04 egonw I would have simply rejected the patch :)
20:04 egonw oh wait...
20:04 egonw it's hidden in a jar
20:04 egonw :)
20:04 egonw oh... or is this at the cdk-signature level?
20:06 maclean No, this is at the cdk level. Otherwise I wouldn't be able to call cdk classes.
20:06 egonw true
20:06 maclean Also, the base signatures have no concept of 'chirality'. Which makes things a bit tricky.
20:06 egonw btw, can you please reply to:
20:06 egonw https://sourceforge.net/tracker/?func=detail&amp​;aid=2998556&amp;group_id=20024&amp;atid=120024
20:06 egonw at some point
20:09 maclean will do
20:20 maclean \o/  [Br]([C@]([Cl][F][I]))
20:20 maclean (test output
20:21 egonw question:
20:21 maclean ?
20:22 egonw chirality at non-root levels will indeed be a horrible pain to read back in
20:22 egonw because of the dept
20:22 egonw the depth may remove ligands of the chiral atom
20:23 egonw so that it becomes impossible to decide what the chirality really means...
20:23 maclean Indeed.
20:23 egonw btw, impressed...
20:23 egonw becuase the @ or @@ depends on the signature string too
20:23 egonw you did that quickly!
20:23 maclean I think that it is undefined if not all atoms are there.
20:23 maclean Hah. I didn't say that it was right yet...
20:23 egonw ah :)
20:23 egonw well still
20:24 egonw maclean++
20:26 maclean well, luckily I had the LigancyFourChiralityTest to copy from :)
20:30 maclean enough for now, anyway. First go pushed.

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