Camelia, the Perl 6 bug

IRC log for #cdk, 2012-03-24

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

All times shown according to UTC.

Time Nick Message
06:20 egonw joined #cdk
08:43 jonalv joined #cdk
09:37 jonalv joined #cdk
09:48 egonw joined #cdk
09:54 egonw I should really find time to write up a tutorial on how to do substructure coloring ... :/
10:46 sneumann joined #cdk
10:55 KevinLawson joined #cdk
10:55 egonw hi Kevin!
10:55 KevinLawson Hi Egon!
10:56 egonw I am really keen on getting your patch working, as it would allow us to close a few bugs in Bioclipse :)
10:56 KevinLawson Great
10:56 egonw KevinLawson: what about making your new code a separate class?
10:56 egonw they have slightly different purpose...
10:57 egonw in that they make different assumptions on the input
10:57 KevinLawson I have been thinking about that and I agree it would be better
10:57 egonw they could share a common interface
10:57 KevinLawson I guess so
10:57 egonw ok, watch this:
10:57 CIA-121 cdk: John May master * r175bb47 / (10 files in 6 dirs): (log message trimmed)
10:57 CIA-121 cdk: Dynamic Settings
10:57 CIA-121 cdk: Provides dynamic setting management on reader/writers and unifies
10:57 CIA-121 cdk: listener code into a single ChemObejctIO implementation
10:57 CIA-121 cdk: New Classes
10:57 CIA-121 cdk: ISetting: interface for which IOSetting now implements (IOSetting is in
10:57 CIA-121 cdk: package 'io' and we needed a way of using it in 'interfaces')
10:57 CIA-121 cdk: John May master * rd5e718a / (17 files in 4 dirs):
10:57 CIA-121 cdk: Implemented new dynamic settings for all usages
10:57 CIA-121 cdk: Change-Id: Ie282ccc381e8f0d959abbdf953266fdab75d1c2b
10:57 CIA-121 cdk: Signed-off-by: Egon Willighagen <egonw@users.sourceforge.net> - http://git.io/uXfuBA
10:58 egonw that's CDK project workflows in action :)
10:58 KevinLawson wow - not really sure quite what is going on here!
10:59 egonw I just pushed two patches from John May (in Chris' group) to the master branch in the GitHub repository
10:59 egonw there is a 'hook' there that send an announcement to an IRC bot, that forwards it to this channel :)
11:00 KevinLawson clever stuff!
11:01 egonw somewhere later today it will get automatically compiled, and reported on here: http://pele.farmbio.uu.se/supernightly/
11:01 egonw there you can always check which tests are known to fail
11:01 egonw for various CDK versions
11:01 KevinLawson OK - that's useful
11:02 egonw indeed
11:03 KevinLawson I have tried out the new algorithm with 1.4.8 and it works fine: Smiles Parser -> fixAromaticBondOrders -> Smiles creator
11:04 egonw good, good
11:05 egonw please let me know all your git questions
11:05 egonw that book I mentioned has a great explanation of rebasing
11:05 egonw that is very helpful in making patches...
11:05 egonw such as merging a series of patches...
11:05 KevinLawson I was also thinking about what you said about detecting the C=O double bonds - of course correcting a bond order between two SP2 hybridised atoms isn't rocket science - but I can see it happening better before the algorithm is called
11:05 egonw brb
11:06 egonw some 10 years ago I worked on two approaches: one bottom up, one top down...
11:06 egonw and just add double bonds to fix valencies...
11:06 egonw however, this was before we had proper atom typing...
11:06 egonw thus, the algorithm would not know if an atom was sp2 or sp3
11:07 egonw causing a ridiculous explosion of options, and it would normally fail on some situations...
11:07 egonw then Todd Martin came along, and gave us this code...
11:07 egonw still before the atom typing...
11:07 egonw my original algorithm would just pick simple atoms first... with only one unknown bond order (terminal, non ring atoms)
11:08 egonw and solve all that first...
11:08 egonw after that, the DBST comes in, to fix the ring systems...
11:08 egonw so, that's what I had in mind to start doing with Klas, when you posted your first patch :)
11:09 egonw ok, need to get another cup of coffe
11:09 KevinLawson fine
11:09 egonw I need to get AMBIT working on my machine today, and preferably get some more work done :)
11:09 egonw for that second, your patch is important :)
11:09 KevinLawson good!
11:09 egonw because that involves Bioclipse, where we are running into trouble with the DBST
11:10 egonw ok, brb
11:11 KevinLawson just for info: I actually got into this because of descriptor calculation in LICSS and the need to see if fixing the bonds made any difference
11:12 KevinLawson I soon discovered that dbst was unworkably slow for a big set of mols
11:12 KevinLawson hence the stuff on splitting it up
11:13 KevinLawson then the realisation that the structure checking algorithm was leading to errors and hence the new direction
11:14 KevinLawson but of course it's really doing something a little different now and a separate class would be good
11:14 egonw you did see that that splitting patch is in 1.4.8, right?
11:15 KevinLawson no actually not!
11:15 KevinLawson this is where I have to come clean - I have not really learned to use the patches - just the main jar
11:15 egonw :)
11:16 egonw see https://github.com/cdk/cdk/commits/cdk-1.4.x
11:16 KevinLawson can you easily get me going with the Unit test system?
11:16 egonw yes
11:16 egonw you got a git checkout, right?
11:16 egonw what IDE are you using? Eclipse?
11:16 KevinLawson ?
11:17 KevinLawson no - NetBeans
11:17 egonw OK, works too, just different...
11:17 egonw I don't have NetBeans, so not sure where the put the unit testing stuff..
11:18 egonw here's an explanation: http://forums.netbeans.org/topic17081.html
11:18 egonw "
11:18 KevinLawson tx
11:18 egonw 1. Have the test file open and press Shift-F6
11:18 egonw 2. Right mouse click on the file and select Run File from the menu
11:18 egonw "
11:19 egonw [03/24/12 12:16] <egonw> you got a git checkout, right?
11:19 KevinLawson not yet
11:19 egonw that was about you having a copy of the full source code repository...
11:19 egonw what windows git tool were you using again?
11:19 KevinLawson no I haven't succeeded in getting it all yet
11:20 egonw ok, hang on...
11:20 * egonw is googling
11:21 egonw http://netbeans.org/kb/docs/ide/git.html
11:21 egonw netbeans 7.1 or better
11:22 KevinLawson I'm on 6.9.1 but I can upgrade
11:22 egonw yeah, probably this is better than installing a separate wingit program
11:23 KevinLawson yes - looks good - I will have a go at sorting this out - got to dash now... : )
11:24 egonw ok, cheers
11:28 KevinLawson thanks, Egon - I will let you know how I get on
11:29 egonw please drop by whenever you like...
11:29 egonw or just to listen what others are talking about :)
11:33 egonw KevinLawson: oh, btw... I think my solution is in fact still around:
11:33 egonw the old code:
11:33 egonw SmilesValencyChecker
11:33 egonw SaturationChecker
11:33 egonw both have .saturate() methods
13:10 jbrefort joined #cdk
19:47 sneumann joined #cdk

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