Camelia, the Perl 6 bug

IRC log for #cdk, 2011-10-10

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

All times shown according to UTC.

Time Nick Message
03:50 sneumann joined #cdk
03:58 egonw joined #cdk
04:58 egonw joined #cdk
05:24 sneumann joined #cdk
05:39 egonw_ joined #cdk
05:43 jbrefort joined #cdk
06:32 egonw_ joined #cdk
06:52 s_wolf joined #cdk
07:04 Gpox joined #cdk
07:17 mgerlich joined #cdk
07:19 mgerlich joined #cdk
07:43 jonalv joined #cdk
09:54 maclean joined #cdk
09:55 maclean morning
11:29 maclean left #cdk
12:01 maclean joined #cdk
12:18 maclean left #cdk
13:07 jonalv I really like this one: http://dl.dropbox.com/u/6280676/serpentine.pdf
13:11 egonw yes, nice
14:43 egonw joined #cdk
14:53 maclean joined #cdk
14:55 maclean hi egonw
14:56 egonw hej maclean
14:56 egonw all fine?
14:56 maclean fine thanks.
14:56 egonw things have been busy here... had little time for the CDK in the past week
14:57 egonw though, I am building something on top of the CDK
14:57 maclean nano-qsar?
14:57 egonw actually... also about to do some work on BODO which affects the CDK
14:57 egonw yes, right
14:57 egonw made good progress with that
14:58 maclean good good.
14:58 maclean I was thinking about the multiple bond system problem over the weekend.
14:58 egonw the BondOrderSystemTool problem?
14:58 maclean (after reading the bioclipse irc logs for friday)
14:58 maclean that's the one.
14:59 egonw the tool is not bad at all
14:59 maclean Oh?
14:59 egonw is it?
14:59 maclean that's good. It has a ring size limitation, though, I think.
14:59 egonw right
14:59 egonw but there are two instrinsic problems, which were simply there at the time:
15:00 maclean the tricky thing would be to use atom types to restrict allowed bond types.
15:00 egonw 1. it's not using CDK atom types, and therefore, cannot take advantage of hybridization info
15:00 egonw 2. it tries to work from scratch and cannot take advantage from info as present in SMILES, such as you know where missing bond orders are
15:01 egonw right
15:01 egonw a long time I wrote both a depth-first and a breadth-first implementation
15:01 egonw but all ran short on the lack of such atom type info
15:01 egonw resulting in trouble for corner cases
15:03 maclean Do the atoms have exact types before the bonds are assigned?
15:04 egonw depends..
15:04 egonw if you start from SMILES, yes
15:04 maclean So, like c1ccccc1
15:04 egonw if you start from PDB, did not do mapping of PDB atom types to CDK atom types, started with 3D, used the RebondTool, then no
15:04 egonw that's a SMILES... here you have six C.sp2 atom types
15:05 egonw moreover, and that's something for master, is that you have bond order info: IBond.Order.UNKNOWN
15:05 maclean That would be very useful.
15:05 egonw or SINGLE_OR_DOUBLE_OR_TRIPLE
15:05 maclean Hmmm.
15:05 egonw yes, we should introduce the UNKNOWN
15:05 maclean Prefer.
15:06 egonw as above, or support IBond.Order == null, like with formal chanrges, etc
15:06 egonw ok, noted
15:06 egonw at some point I started a patch
15:06 egonw but
15:06 egonw the effects were massive :)
15:06 maclean Heh.
15:06 egonw so much code to update and generalize ...
15:06 egonw but we have master now :)
15:06 maclean True.
15:07 egonw I'm tempted to even allow such a patch, breaking ton of stuff...
15:07 egonw then we have a good incentive to start working...
15:07 egonw master is too silent right now
15:07 egonw we need more havoc
15:08 egonw (e.g. we also need to 'fix' the lack of UNKNOWN for IS_AROMATIC)
15:08 egonw if you query for that right now, it's NOT aromatic
15:08 egonw since we use a boolean[] for flags
15:09 maclean IS_AROMATIC?
15:09 maclean 'query' as in QueryAtomContainer?
15:10 egonw no, just the flag which we use for IAtom and IBond to indicate if the atom or bond is aromatic
15:10 jonalv egonw: let's provide some havoc on sunday then :)
15:11 egonw jonalv: sounds good!
15:11 egonw jonalv: yeah, we should not forget to make an agenda
15:11 jonalv egonw: well, if we can apply my commits to cdk master that would be good + probably create some havoc if that is a good thing...
15:12 jonalv :)
15:13 maclean when on sunday?
15:13 jonalv good question
15:14 egonw during the day, if there are no other plans...no confirmation yet
15:14 jonalv I suppose where is at least for me at Egons place?
15:14 jonalv ah
15:14 jonalv true that
15:14 jonalv it's not confirmed
15:15 maclean ah ok.
15:17 maclean ok, one last question egonw? is this :
15:17 maclean http://www.daylight.com/daycgi/d​epict?4f53283d4f29284f29283d4f29
15:17 maclean the 'same' as :http://www.daylight.com/daycgi/d​epict?4f53283d4f29283d4f29284f29
15:20 egonw_ joined #cdk
15:20 egonw_ jonalv: yeah, we should not forget to make an agenda
15:21 jonalv egonw: what do you have in mind?
15:21 egonw to take your PhD work as a starting point
15:23 jonalv I can provide a goal for sunday. To get this branch: https://github.com/jonalv/cdk ready for CDK master
15:23 jonalv but I amk not sure if it is feasable
15:23 egonw sounds good to me
15:24 jonalv I suppose this one: https://github.com/jonalv/cdk/tree/KlekotaRoth into 2.something could be another goal
15:24 jonalv or is that
15:24 jonalv 1.something
15:24 jonalv I am confused now
15:25 jonalv it's 1.4 right?
15:26 egonw KlekotaRoth should be aimed at 1.4 indeed
16:04 sneumann joined #cdk
16:56 egonw joined #cdk
18:13 sneumann joined #cdk
18:43 jbrefort joined #cdk

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