Camelia, the Perl 6 bug

IRC log for #cdk, 2012-10-18

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

All times shown according to UTC.

Time Nick Message
06:25 konditorn joined #cdk
06:41 egonw joined #cdk
06:58 sneumann joined #cdk
07:02 Gpox joined #cdk
07:33 jonalv joined #cdk
07:40 konditorn egonw: did you know that the clone-method in AtomContainer don't return an identical object?
07:40 konditorn i.e. molecule.equals(molecule.clone()) == false.
07:51 konditorn …even if 'molecule' is an empty  AtomContainer...
08:36 jbrefort joined #cdk
09:01 konditorn egonw: ping
09:18 egonw pong
09:19 egonw .equals() should be based on the hashcode...
09:19 egonw they *are* different object, so equals should not match
09:19 egonw IIRC
09:23 konditorn isn't equals used to compare two different objects to se if they is equal, even if the reference isn't the same?
09:23 egonw ask Jonathan and Arvid... they may know better...
09:23 konditorn http://docs.oracle.com/javase/6/do​cs/api/java/lang/Object.html#clone()
09:23 egonw equals() is not very well worked out in the CDK
09:23 egonw (as in: undefined)
09:24 egonw there are questions like: are flags, properties, part of what something is... and if those should be reflected in the hashcode, and thus equals...
09:25 konditorn the basic problem is the InChI: if i clone an AtomContainer I don't get the same InChI form the clone as from the "real"...
09:37 konditorn afk: food.
09:37 egonw ah, that is a big problem...
09:37 egonw it means that crucial information is list...
09:38 egonw after lunch, give me an example, so that I can see what information is lost...
09:38 egonw quick guess (and I feel embarished...)
09:38 egonw it's probably stereochemistry lost ?
10:51 jonalv yea that's what we were thinking too
10:52 jonalv ah regarding equals and hash code… If you change one you need to change the other… However, not messing with the is of course a way as well. Seems to be what CDK does?
10:55 konditorn as jonalv said, we think it's the stereochemistry. I'll try to write some test for it.
10:56 konditorn the warning message that I get from the InChI-generator is:
10:56 konditorn WARNING: Wrong 0D stereo descriptor(s): #1; #2; #3; #4; Omitted undefined stereo
10:57 konditorn and the molecule I used is our friend from bug 3170...
11:07 egonw what are the two InChIs you get, before and after cloning?
11:21 jonalv_ joined #cdk
11:21 Gpox joined #cdk
11:23 konditorn before cloning: …13-14H2,1-6H3[/t19-,20+,23+,24-]
11:23 konditorn and after: 13-14H2,1-6H3[]
11:26 Gpox joined #cdk
11:27 jonalv joined #cdk
11:45 cruttkie joined #cdk
16:00 jbrefort joined #cdk
18:54 sneumann joined #cdk
20:14 jbrefort joined #cdk

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