Camelia, the Perl 6 bug

IRC log for #cdk, 2010-08-08

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

All times shown according to UTC.

Time Nick Message
03:11 azeem joined #cdk
05:43 jbrefort joined #cdk
06:25 s9asad joined #cdk
08:04 bag_ joined #cdk
08:35 egonw joined #cdk
10:10 maclean joined #cdk
11:20 s9asad joined #cdk
13:20 egonw s9asad: left my comments in the tracker
13:20 egonw brb
14:00 maclean joined #cdk
14:00 egonw hi maclean
14:00 maclean hi egonw
14:00 egonw maclean: you saw the permutation code in the SMSD patch?
14:01 maclean Mayybee.
14:01 maclean Depends whether you want me to write tests for that code, or not...
14:01 maclean :)
14:01 egonw ummm... it's on the SMSD patch...
14:02 egonw it's Asad's code
14:02 egonw so, why should you write test code?
14:02 maclean Heh. Yes, yes. that's right. It IS Asad's code. So he should definitely write tests for it.
14:02 maclean Ok, I admit. It might actually be my code.
14:02 egonw please leave as comment... I had not spotted the missing unit tests...
14:02 egonw it's a MF
14:03 egonw :)
14:03 egonw I could not tell
14:03 egonw the code did not have copyright/license headers...
14:03 egonw that I did report as MF
14:03 maclean Well, the lack of headers is my fault.
14:04 maclean It's from that repository I linked to on Friday.
14:04 maclean Oh, and from that discussion : no, you can't use morgan numbers to canonically label, as they only do a symmetry partition, not an automorphic partition. Or something.
14:05 egonw true
14:05 maclean Anyway, eg: all benzene carbons will have the same morgan number.
14:05 egonw as symmetrical atoms will end up with the same number, right?
14:05 egonw right
14:05 egonw that could explain why we have that canon SMILES bug report
14:05 egonw where the SMILES is *not* canon
14:06 maclean Obviously that symmetry partition could then feed into another technique that does do automorphism partitioning, but I haven't put that code as a CDK patch yet...
14:06 maclean which canon smiles bug report?
14:07 egonw bug tracker?
14:07 egonw or ML?
14:07 egonw dunno
14:07 maclean ok, will look on bug tracker.
14:07 * egonw is writing more pharmbio.org course material
14:10 maclean "Rule Based Learning and Rough Sets" :)
14:10 maclean Heh hehe.
14:26 maclean egonw : the SmilesGenerator method parseChain seems to have a lot of code that calculates bond angles when creating chiral smiles - this is 2D, right?
14:26 maclean And what about if the molecule has no 2D coords?
14:27 egonw dunno
14:27 egonw I did not write that code
14:27 maclean well, I know who did...
14:27 egonw I think I do too
14:27 egonw it's wedge bond based, not?
14:28 egonw if so, no matter who had written it, I would not trust it... too complex and undefined a problem, with the much bad input
14:28 maclean It is, yes. But there are calls to BondTools.giveAngle
14:29 egonw which makes sense too then...
14:29 egonw as you need to know where the other bonds are pointing, to decide what a wedge means
14:30 maclean Yes, I suppose.
14:30 maclean But if it is 2D, this is all for a molecule embedded in a plane.
14:30 maclean Which works, I suppose.
14:31 egonw well, under the right circumstances it does :)
14:31 maclean Well, anyway, would be nice if it was factored out into a separate class. Yet another XTool(s) class, probably.
14:31 egonw indeed
14:36 * maclean is off to sit in the garden perhaps
17:27 s9asad joined #cdk
19:08 maclean joined #cdk
20:15 sneumann_ joined #cdk
20:22 sneumann_ joined #cdk
20:33 sneumann_ joined #cdk
22:52 bag_ joined #cdk
23:12 s9asad joined #cdk

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