Camelia, the Perl 6 bug

IRC log for #cdk, 2010-06-03

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

All times shown according to UTC.

Time Nick Message
02:28 s_wolf joined #cdk
05:25 sneumann joined #cdk
05:46 egonw joined #cdk
05:47 egonw joined #cdk
06:19 egonw joined #cdk
06:53 Gpox joined #cdk
07:04 egonw joined #cdk
07:21 sneumann joined #cdk
08:42 sneumann Wow, this looks like fun:
08:42 sneumann http://online.wsj.com/article/SB100014​24052748703341904575266583403844888.ht​ml?mod=WSJ_business_IndustryNews_DHC
08:42 zarah sneumann's link is also http://tinyurl.com/3596jnn
08:42 sneumann Glaxo Tries a Linux Approach
08:42 sneumann Drug Maker Shares Its Research Data Online in Test of Open-Source Principles
08:44 jbrefort joined #cdk
09:00 egonw CC0 is like the MIT or BSD license, but then for data
09:00 egonw and not actually a license, but a legal waiver
09:10 bag_ joined #cdk
09:37 egonw joined #cdk
09:41 egonw_ joined #cdk
09:52 bag_ joined #cdk
11:11 egonw joined #cdk
13:03 carsten joined #cdk
13:58 CIA-47 cdk: mark_rynbeek * r15550 /jchempaint/trunk/src/main/org/op​enscience/jchempaint/controller/ (Rotate3DModule.java RotateModule.java): Speeds up rotation
14:31 maclean joined #cdk
14:31 maclean hi
14:31 zarah ni hao maclean
14:37 egonw hej maclean!
14:37 maclean hej
14:37 * egonw is closing in on finishing the rendering parameter API refactoring
14:37 maclean cool.
14:38 egonw RendererModel is much smaller already
14:38 maclean excellent news.
14:38 maclean can I help?
14:38 egonw 10 params left
14:38 egonw you could, if you like
14:39 egonw would be very much appreciated actually!
14:39 maclean it might be more effort for you though, not less :) Perhaps I should do the javadocs on the classes I wrote (eg rendering visitors).
14:39 egonw ummm... let me upload what I have right now...
14:39 egonw yes, that would be a good idea too
14:39 egonw it's not that difficult though
14:40 egonw but like all refactoring, it needs to be done with care... so it takes time
14:40 maclean true. I think of it as 'code surgery' sometimes.
14:41 egonw indeed
14:41 egonw often with collateral damage, more than in surgery, I hope :)
14:42 egonw I think the function and role of the various part of the human body do not need JavaDoc :)
14:43 maclean :)
14:45 sneumann Query OK, 2041801 rows affected (5 min 31.94 sec)
14:45 sneumann '/tmp/metfragLongs.csv'
14:47 sneumann sorry, wrong channel
14:47 sneumann egon: that was the number of pubchem smiles and inchi which don't fit into 200 chars ;-)
14:48 egonw :)
14:48 egonw about 7% or so?
14:51 sneumann Yup. Only 62.561 SMILES exceed 198 chars, but 2.0 mio Inchis exceed 198 chars
15:06 egonw shit
15:06 egonw cdk-jchempaint-10 does not compile ...
15:06 slyrus I thought Inchis were supposed to be short!
15:06 egonw that's inchikey
15:06 slyrus oh, right
15:07 sneumann :-)
15:09 maclean joined #cdk
15:27 egonw maclean: don't update from pele just yet
15:27 maclean ok
16:54 egonw maclean: I'm uploading CDK-JChemPaint 10 now
16:54 maclean ah great.
16:55 egonw tag pushed to ssh://pele.farmbio.uu.se/var​/www/git/cdk-jchempaint.git
16:55 egonw 10 rendering parameters left to refactor
17:04 egonw ok, stuff is on SF too now
17:15 sneumann joined #cdk
17:22 slyrus egonw: does this mean that there's a jchempaint that I should be able to build now, or is this just the cdk-that-goes-with-jcp?
17:27 maclean slyrus : there is a Jchempaint application/applet, and there is the CDK rendering core. It's all a bit confusing at the moment...
17:28 slyrus does the CDK rendering core allow me to draw a 2d depiction of a molecule on, say, a canvas?
17:29 maclean It should, yes :)
17:29 maclean Either scaled-to-fit or at a constant bond length.
17:31 maclean There are details on egon's blogh http://chem-bla-ics.blogspot.com/2010/04/​cdk-jchempaint-4-embedding-renderer.html
17:31 zarah maclean's link is also http://tinyurl.com/3xx7dhg
17:32 maclean or there is a tutorial on the wiki, that is mostly the same, but a bit older : http://sourceforge.net/apps/mediawiki/cdk/inde​x.php?title=Renderer_Tutorial-Single_Molecules
17:32 zarah maclean's link is also http://tinyurl.com/342rpvs
17:33 egonw version 10 fixes the does-not-respond-to-new-parameter-values bug
17:33 egonw moreover, Chris and I are going to try to plan a new JCP workshop to integrate more of the applet code into the rendering core branch
17:34 maclean that would be good.
17:34 maclean this was interesting:
17:34 slyrus awesome, thanks!
17:34 maclean http://verpa.wordpress.com/2010/02/12/ho​w-to-add-a-feature-to-jchempaint-part-2
17:34 zarah maclean's link is also http://tinyurl.com/3728mgu
17:34 slyrus I'll try this tonight
17:35 maclean (blog post about adding a dialog to JCP to specify the rotation exatly).
17:36 egonw mmm... can't say I remember that post
17:39 egonw ok, let's see if I can get CDK 1.3.5 and CDK-JChemPaint to export to Eclipse plugins, and update Bioclipse 2.4.x ...
17:46 maclean hmmm. copy/pasting a 1Mb patch file into vi wasn't the greatest of ideas.
17:46 egonw :)
17:46 egonw you're quite git literate, not?
17:46 egonw I suggest using the git repos
17:49 maclean well, I tried doing that, then discovered I had forgotten how to access the repo or didn't know the right url.
17:49 maclean the patchfile doesn't seem to apply, anyway.
17:50 egonw oh... really?
17:50 egonw not good...
17:50 egonw it's made with: git diff cdk-1.3.5.. > cdk-jchempaint-10.patch
17:50 egonw what's the error you get?
17:51 maclean error: patch failed: .classpath:20
17:51 maclean error: .classpath: patch does not apply
17:51 maclean error: patch failed: build.props:1
17:51 maclean error: build.props: patch does not apply
17:51 maclean error: patch failed: build.xml:572
17:51 maclean error: build.xml: patch does not apply
17:51 maclean error: cannot apply binary patch to 'jar/swt-gtk-3.4.2.jar' without full index line
17:51 maclean error: jar/swt-gtk-3.4.2.jar: patch does not apply
17:51 maclean error: patch failed: src/main/org/openscience/cdk/con​troller/IMouseEventRelay.java:1
17:51 maclean error: src/main/org/openscience/cdk/co​ntroller/IMouseEventRelay.java: patch does not apply
17:51 maclean ...etc
17:51 egonw mmm...
17:51 maclean the jar one looks weird.
17:53 carsten joined #cdk
17:54 egonw this works for me:
17:54 egonw git checkout -b 157-testPatch cdk-1.3.5
17:54 egonw (pick whatever branch name you like)
17:54 egonw patch -p 1 < ../cdk-jchempaint/cdk-jchempaint-10.patch
17:54 egonw note the '-p 1'
17:56 maclean hmmm. ok.
17:57 maclean I was using git apply, as patch -p0 didn't work.
17:57 maclean yes, -p1 works, but with some failed hunks.
17:58 egonw it applies cleanly here...
17:58 egonw line break problem?
17:58 maclean applying to the wrong version problem, I expect.
18:00 maclean Line breaks doesn't seem likely, as I am on OSX.
18:04 egonw ah, wrong version can certainly be the reason :)
18:05 egonw grmph...
18:05 egonw the standard inchi patch is not reviewed yet, so not in 1.3.5
18:06 egonw I though it was ... :(
18:06 egonw https://sourceforge.net/tracker/?func=detail&amp​;aid=3005890&amp;group_id=20024&amp;atid=320024
18:12 maclean alright. I give up on patch files. What is the pele url? Pulling from the ssh: required login details.
18:14 egonw mom
18:14 maclean doing stuff can be fracking hard sometimes.
18:14 egonw need to check how git update-server-info works with Git 1.6
18:15 egonw ok, try this URI:
18:15 maclean ahhh. I did get some update-refs complaint at one point.
18:15 egonw http://pele.farmbio.uu.se/git/cdk-jchempaint.git
18:15 zarah egonw's link is also http://tinyurl.com/2ut9nha
18:16 egonw I've updated the refs now
18:18 * maclean discovers git clean
18:18 egonw -fd for force and directories
18:18 maclean I swear it was hidden away in a jungle temple full of traps, not just clearly visible in any tutorial....
18:19 egonw :)
18:28 egonw there we go...
18:28 egonw well, at least the bioclipse 2.4 copy
18:28 CIA-47 org.openscience.cdk: Egon Willighagen bioclipse2.4 * r740d1cd / (537 files in 177 dirs): Uploaded CDK 1.3.5 + Standard InChI patch + CDK-JChemPaint 10 - http://bit.ly/bzgPiR
18:29 egonw mom, got to reboot konversation
18:29 egonw joined #cdk
18:33 maclean egonw : so when I pull, it claims to be up-to-date already. :( 'cdk-jchempaint-10' is a tag, right?
18:33 egonw yes
18:34 egonw the matching hash should be:
18:34 egonw 465d556c95ab624a2136b259187a14964b174a59
18:36 maclean ahhhh!
18:38 CIA-47 org.openscience.cdk: Arvid Berg bioclipse2.4 * rd7fe08f / (14 files in 14 dirs): Removed net.bioclipse.xom from Require-Bundle - http://bit.ly/bfsjro
18:38 CIA-47 org.openscience.cdk: Arvid Berg bioclipse2.4 * ra01983d / (4 files in 4 dirs): Removed org.xmlcml.xmlxom from Require-Bundle - http://bit.ly/bHdawM
18:38 CIA-47 org.openscience.cdk: Arvid Berg bioclipse2.4 * r2088c56 / (3 files in 3 dirs): Added Import-Package nu.xom - http://bit.ly/9577JX
18:38 CIA-47 org.openscience.cdk: Arvid Berg bioclipse2.4 * r2c425df / (4 files in 4 dirs): Added Import-Package nu.xom and org.xmlcml.cml.* - http://bit.ly/akup2x
18:38 CIA-47 org.openscience.cdk: Egon Willighagen bioclipse2.4 * r70294c9 / (5 files in 5 dirs): Just depend on nu.xom; XOM 1.2 is fine too... can we depend on verion=1.x in some way? - http://bit.ly/bgiIIU
18:41 CIA-47 org.openscience.cdk: Egon Willighagen bioclipse2.4 * r880dbce / (2 files in 2 dirs): Removed qsar.model.data package with only data, no java classes - http://bit.ly/cBzc24
18:43 egonw only 58 compile errors in Bioclipse 2.4.x
18:47 maclean Ok, so I got it to work, but only by committing a tiny change to build.props.
18:47 egonw ok, that's not uncommon
18:47 maclean I know that is a terrible idea, however...
18:47 egonw that file is so small, that almost anything conflicts
18:48 egonw just commit it in a different patch
18:48 maclean ok
18:53 maclean hmmm. also, it needs doccheck.jar to be removed (as usual) to build.
18:53 egonw yeah, another unreviewed patch... :(
18:54 maclean Ah, I see.
18:54 maclean Well, basically, Rajarshi does all the reviewing, right :)
18:54 egonw unless he wrote the patch :)
18:54 egonw yeah, we seriously need more reviewers
18:57 maclean Hmmm. I see a problem with the BasicGenerator. It doesn't implement IGenerator, and therefore doesn't register the parameters of its 'contained' generators.
18:59 egonw there is a split up of IAtomContainerGenerator and IChemModelGenerator
18:59 egonw had to do that to fix some dependencies...
19:00 maclean Yah, but it doesn't implement either :) I think I wrote BasicGenerator, so it's my fault...
19:00 egonw oh... BasicGenerator
19:00 egonw not Basic*Generator
19:00 egonw ic...
19:00 egonw well, not really
19:01 egonw I had *not* seen that class yet :)
19:01 maclean :) It's just a convenience class.
19:02 egonw good one too...
19:02 maclean BasicSceneGenerator is weird - it makes nothing, but is the home of many parameters.
19:03 maclean I think it means that you always have to have one in the generator list.
19:03 maclean Ah... and indeed it is used in the AWTDrawVisitor.
19:03 egonw the first line is 100% correct
19:03 egonw it might have place to draw grids or so...
19:04 egonw dunno
19:04 egonw but I had to put those params somewhere...
19:04 egonw not pretty, but could not htink of sth better
19:04 egonw 35 compile errors left
19:07 egonw btw, there is a serious problem with the code...
19:07 egonw getScale() is all over the place...
19:07 maclean yup.
19:07 egonw that is supposed to be hidden as affine transformation...
19:07 maclean no no no.
19:07 maclean also : no.
19:07 egonw why not again?
19:08 egonw why should the generators not draw in world space, and let affine transformations fo the rest
19:08 egonw ?
19:09 maclean I can't even remember anymore.
19:09 maclean Originally, it was due to a mac bug.
19:10 maclean Also, because I wanted dictatorial control over how the diagram scales and is drawn.
19:10 maclean But I'm sure there were better, more defensible reasons.
19:10 egonw :)
19:11 egonw yeah, the scale all over the code was one of the really bad things about the old code...
19:11 egonw I should have paid more attention that it did not sneak back in...
19:13 maclean What was wrong with model.getScale(), by the way?
19:13 maclean Is the intent to get rid of RenderModel totally?
19:13 maclean Because a small default model seems reasonable.
19:15 egonw that's a parameter now
19:16 maclean Well yes, I can see.
19:21 bag_ joined #cdk
20:28 jbrefort joined #cdk

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