Camelia, the Perl 6 bug

IRC log for #cdk, 2009-03-17

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

All times shown according to UTC.

Time Nick Message
03:50 bag joined #cdk
03:50 goomba joined #cdk
03:50 Conrad joined #cdk
03:50 CIA-62 joined #cdk
06:05 jbrefort joined #cdk
06:15 carsten joined #cdk
07:35 Gpoks joined #cdk
09:01 shk3 joined #cdk
09:29 sneumann joined #cdk
10:06 egonw joined #cdk
11:40 egonw joined #cdk
11:40 steinbeck joined #cdk
13:13 egonw joined #cdk
13:22 egonw joined #cdk
13:24 egonw joined #cdk
13:27 Gpox joined #cdk
13:48 CIA-62 cdk: rajarshi * r14504 /nightly/trunk/nightly.config.sample: Added informative comment
13:48 rajarshi joined #cdk
13:50 rajarshi good morning
13:59 egonw joined #cdk
14:01 egonw hi rajarshi
14:04 rajarshi hi egonw
14:13 egonw rajarshi: what is SMARTS for AnyAtom ?
14:13 rajarshi *
14:13 rajarshi e.g.: C*C
14:14 egonw ah, no square brackets
14:14 egonw thanx
14:15 rajarshi no
15:15 shk3 egonw: how can I see the change of a git commit?
15:16 shk3 last commit messages I get in the code browser on sf or two days ago
15:18 rajarshi git diff/
15:19 egonw_ joined #cdk
15:31 egonw rajarshi: can I interest you in setting up FP prescreening for the 2D template library?
15:31 egonw so that we can use templates even when symbols and bond orders do not match?
15:32 rajarshi can you explain?
15:32 egonw right now, the 2D templates needs an exact match
15:32 rajarshi AnyAtomAnyBond query?
15:32 egonw yes
15:33 egonw but that makes it very, very slow
15:33 egonw tried that...
15:33 egonw and you get time outs really fast...
15:33 egonw so we need some prescreening
15:33 rajarshi ok
15:33 rajarshi so why not just convert templates to carbon only structures
15:33 egonw to have some indication that the 2D template can be subscructure
15:33 egonw they are
15:33 egonw carbon only, single bond only
15:34 rajarshi so why the slowdown/problem?
15:34 egonw because templates match too good
15:34 egonw before they fail
15:34 rajarshi aah ok
15:34 rajarshi so basically first compute fp, see if it's a possible match then do actual isomorphism
15:35 egonw yes...
15:35 egonw that was what I was thinking
15:35 rajarshi doesn't that already happen? Why do we generate the fp's?
15:35 egonw that's 3D
15:35 egonw this is 2D
15:35 rajarshi aah, OK
15:35 egonw we actually do not have so many templates
15:35 rajarshi how many?
15:36 egonw but with anyatomanybond, matching just takes very long
15:36 egonw 4-5 or so
15:36 egonw will be more
15:36 egonw but since they currently only apply when strictly matching...
15:36 egonw it does not make much sense
15:36 rajarshi where is the template matching done?
15:36 egonw e.g. sterane skeleton should match no matter where a double bond is...
15:37 rajarshi right
15:37 egonw TemplateHandler
15:40 rajarshi if this plan goes ahead is there a reson to keep mapTemplateExact?
15:40 egonw not sure how that is used
15:41 egonw is it actually used?
15:41 rajarshi so the main entry point is mapTemplates/
15:41 rajarshi it's used in SDG, line 511
15:42 egonw weird... does not show up here...
15:43 rajarshi hmm, but going to fp screening implies that the code may be bottlenecked on the fp calculation
15:44 egonw true
15:44 rajarshi anyway, fp screening should be easy to include
15:44 egonw maybe just count ring sizes then?
15:45 rajarshi i think fp screening bottleneck is more minor compare to exact ss matching
15:45 rajarshi worst case, it will timeout on bcomplex rings
15:45 rajarshi we can see what happens with fp screen
15:51 carsten joined #cdk
16:05 rajarshi egon, when doing template matching does the query molecule come in with aromaticity detected and atom typed? Should this be done to the templates as well?
16:05 egonw the templates do not have those concepts
16:05 egonw not when anyatomanybond
16:06 rajarshi oh ok
16:06 rajarshi this implies that query mols are also any atom any bond right?
16:07 egonw no, they are not
16:07 egonw afaik
16:07 rajarshi that will be a problem in fp screening then
16:07 egonw but the template is being tested as substructure, and thus the IQueryAtomCotnainer
16:07 rajarshi C1CCN1 will not set the same bit as C1CCC1
16:07 egonw right
16:07 egonw forgot to mention that
16:08 egonw yes, we'll need the graphonlyFP
16:08 rajarshi does that do rings/
16:08 egonw no, it's path only
16:09 rajarshi ok, well we can always change the choice of fp easily
16:09 egonw right
16:26 rajarshi egon, how can I read a CML file when I have the filename?
16:33 egonw FileInputStream
16:35 rajarshi thanks, solvd
16:35 rajarshi damn, Fingerprinter does atom type perception ?
16:37 egonw the main Fingerprinter might actually
16:37 egonw since it uses aromaticity
16:37 egonw sorry, forgot about that...
16:37 rajarshi not a big issue
16:37 rajarshi but a pain running the fp generator from the command line
16:48 zarah joined #cdk
16:53 egonw going to catch my bus now
16:53 egonw bbl
16:53 rajarshi bye
18:40 egonw joined #cdk
18:43 rajarshi hi
18:43 zarah hi rajarshi
18:48 egonw hej
18:53 rajarshi 2t template fp's done, patch put into the tracker
18:53 egonw ha, a patch...
18:53 egonw yes, just saw it...
18:53 egonw did you measure a speed up?
18:54 rajarshi slowdown for smaller mols and a slight speedup for larger mols
18:54 rajarshi but slowdown is on th eorder of 10's of ms
18:54 egonw and the speed up?
18:54 rajarshi 10's of ms as well
18:55 rajarshi but depends on the molecule
18:55 egonw and I guess on the size of the template library
18:55 rajarshi yes
18:55 egonw good... time to make more templates :)
18:55 rajarshi we'll see a better payoff for larger templaet libs
18:55 rajarshi btw, what's the time scale for inclusion of patches
18:55 rajarshi I won't be running this repo after 3 or 4 weeks
18:55 egonw the time people have to review...
18:56 egonw we'll see...
18:56 egonw I can easily copy them to pele when needed
18:56 rajarshi do we still need 2 reviews for trunk?
18:56 egonw no, one review for trunk should be OK
18:57 rajarshi ok - all except the PT stuff are qiute small patches
19:07 bag joined #cdk
19:09 egonw rajarshi: I am writing a bot that will monitor the git RSS feed
19:09 egonw and post items when they appear
19:09 rajarshi nice
19:09 rajarshi replaces CIA then?
19:10 egonw yes
19:11 egonw SF does not seem to have git hooks
19:11 egonw so no email either...
19:11 egonw RSS is a good replacement for tha
19:11 egonw +t
19:11 egonw only CIA was really missing...
19:11 egonw I hope this code fixes that
19:13 timvdm joined #cdk
19:19 rajarshi sounds good
19:22 rajarshi egon, how do you write such a bot?
19:22 egonw mom...
19:23 egonw when it works, i'll put the source code on GitHub
19:24 rajarshi ok
19:36 rednael joined #cdk
19:40 egonw so, let's see
19:41 rednael Added generics
19:41 rednael Tue Mar 17 20:41:03 GMT+01:00 2009
19:41 rednael Egon Willighagen <egonw@users.sourceforge.net>
19:41 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=a49d5​59be970f83e7aa35a9d78861e87c88d303b
19:41 rednael Removed an unused import
19:41 rednael Tue Mar 17 20:39:51 GMT+01:00 2009
19:41 rednael Egon Willighagen <egonw@users.sourceforge.net>
19:41 zarah rednael's link is also http://tinyurl.com/c5p9vo
19:41 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=4a888​f0a7b47423c971cc047fbea25a2ff54a256
19:41 zarah rednael's link is also http://tinyurl.com/dyv5wo
19:41 rajarshi nice
19:41 rajarshi can you modify zarah to ignore links from rednael?
19:41 rajarshi (what does rednael mean?)
19:42 egonw @tell masak please have zarah ignore links to cdk.git.sourceforge.net made by rednael
19:42 zarah Consider it noted.
19:42 egonw rajarshi: zarah leander was some prewar babe
19:42 egonw one of the first google hits on 'zarah'
19:43 egonw leander nick was already registered, so reversed it
19:43 rajarshi ha ha
19:43 egonw she was swedish
19:44 rajarshi were these commits to sf master?
19:44 egonw ah... good one...
19:44 egonw no, cdk-1.2.x
19:44 rajarshi ok, it'd be useful to somehow make that clear
19:44 egonw yes, enough to improve :)
19:52 egonw git clone git://github.com/egonw/rednael.git
19:54 rajarshi pretty cool
19:59 rednael joined #cdk
20:00 egonw OK, that should do it
20:00 egonw will watch some tv now
20:00 egonw but bbl
20:00 rajarshi later
20:01 egonw feel free to play with the code :)
20:01 egonw just fork :)
20:38 jpansanel moin
20:38 zarah saluton jpansanel
20:40 rajarshi hi
20:40 zarah oh hai rajarshi
21:08 egonw hi jpansanel
21:10 jbrefort joined #cdk
21:16 rednael2 joined #cdk
21:21 rednael2 joined #cdk
21:21 egonw right
21:26 rednael2 joined #cdk
21:41 rajarshi of for the day, good night
21:41 egonw bye
21:55 rednael Use interface instead of implementation
21:55 rednael Tue Mar 17 22:55:12 CET 2009
21:55 rednael Egon Willighagen <egonw@users.sourceforge.net>
21:55 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=a51dc​cb73aa52586077df195b112d713588db6ed
21:55 rednael Added generics
21:55 rednael Tue Mar 17 22:49:21 CET 2009
21:55 rednael Egon Willighagen <egonw@users.sourceforge.net>
21:55 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=12242​ccefa877baa6a167e2d9fa86bf011b9a973
21:55 zarah rednael's link is also http://tinyurl.com/c78ehg
21:55 zarah rednael's link is also http://tinyurl.com/d6wy2k
22:01 rednael Added generics, here too
22:01 rednael Tue Mar 17 23:01:23 CET 2009
22:01 rednael Egon Willighagen <egonw@users.sourceforge.net>
22:01 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=3f17e​5f5743741181b3bd6d1dc6e2161aeca8312
22:01 zarah rednael's link is also http://tinyurl.com/dn4tpc
22:11 rednael Some other code cleanup
22:11 rednael Tue Mar 17 23:10:58 CET 2009
22:11 rednael Egon Willighagen <egonw@users.sourceforge.net>
22:11 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=81e9e​2d0143b2ac74838d1434fb86efd4477b1d2
22:11 zarah rednael's link is also http://tinyurl.com/cxd8av
22:24 rednael Added generics, one last time... Leander should work by now...
22:24 rednael Tue Mar 17 23:23:57 CET 2009
22:24 rednael Egon Willighagen <egonw@users.sourceforge.net>
22:24 rednael http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=2edf6​e45816022b23604bfe9ab431cd3e261de8c
22:24 zarah rednael's link is also http://tinyurl.com/cj22bs
22:30 rednael joined #cdk
22:33 rednael [cdk-1.2.x] Added generics  http://cdk.git.sourceforge.net/git/g​itweb.cgi?p=cdk;a=commitdiff;h=c5565​5ebe97fe9e9a489ad2faf95431fbafce142
22:33 zarah rednael's link is also http://tinyurl.com/cep4lu

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