Camelia, the Perl 6 bug

IRC log for #cdk, 2009-04-30

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

All times shown according to UTC.

Time Nick Message
04:09 bag joined #cdk
05:38 jbrefort joined #cdk
06:09 egonw joined #cdk
06:27 Andreas joined #cdk
07:14 Gpox joined #cdk
07:15 egonw hi Gpox
07:15 Gpox hi egonw
07:46 Andreas hi
07:46 zarah oh hai Andreas
07:48 sneumann Morning!
07:49 sneumann Does anyone know an "official" location for the CML schema description
07:49 sneumann http://msbi.ipb-halle.de/~sneumann/cmlschema.html
07:49 zarah sneumann's link is also http://tinyurl.com/dcrofc
07:49 sneumann which came from a CML paper supplementary
07:49 sneumann ?
08:05 egonw I am not really aware of that...
08:05 egonw the 2.5b1 schema in CDK SVN seems to be the definate one...
08:27 sneumann Path ?
08:28 sneumann Does it have the HTML documentation, or just the xsd ?
09:25 CIA-58 cdk: goglepox * r14578 /cdk/branches/jchempaint-primary/src/​main/org/openscience/cdk/controller/ (3 files in 2 dirs):
09:25 CIA-58 cdk: Changed from int[] to Integer[] array for hydrogencount in undo
09:25 CIA-58 cdk: This enables it to store the null vales and avoids NPEs
09:31 shk3 joined #cdk
09:39 maclean joined #cdk
10:31 sneumann git behind proxy ?
10:31 sneumann I tried pulling cdk via
10:31 sneumann git clone git://cdk.git.sourceforge.net/gitroot/cdk
10:31 sneumann and
10:32 sneumann git clone http://cdk.git.sourceforge.net/gitroot/cdk
10:32 zarah sneumann's link is also http://tinyurl.com/cqwbf7
10:32 sneumann without success.
10:32 sneumann Any suggestions short of contacting the net admins ?
10:33 maclean what happens when you try to clone?
10:33 maclean I mean; how does it fail?
10:34 sneumann for git:// it is: cdk.git.sourceforge.net[0: 216.34.181.91]: errno=Connection timed out
10:35 sneumann for http:// it is
10:35 sneumann Cannot get remote repository information.
10:35 sneumann Perhaps git-update-server-info needs to be run there?
10:39 maclean hmmm. I've just tried with "git clone git://cdk.git.sourceforge.net/gitroot/cdk", on my machine, and it starts off okay
10:39 maclean perhaps some firewall issues?
10:39 sneumann Yes, git:// uses port 9418
10:39 sneumann which is probably blocked here
10:40 sneumann workaround would be http access
10:40 sneumann to the git repository
10:40 maclean right, I see
10:41 maclean ah, I get "fatal: http://cdk.git.sourceforge​.net/gitroot/cdk/info/refs not found: did you run git update-server-info on the server?"
10:41 zarah maclean's link is also http://tinyurl.com/cpuj98
10:41 maclean when I try with http.
10:41 sneumann yup
10:43 maclean can you use ssh?
10:43 maclean hmmm, maybe that would require write permissions
10:43 maclean I guess only egon has that, I don't know.
10:44 sneumann I dont even want to be able to write ;-)
10:44 maclean no, I know :)
10:44 maclean sorry, I'm not much help here, I guess egon's the guy to ask, as always...
10:48 sneumann is it possible I need another url, because http is there
10:48 sneumann http://cdk.git.sourceforge​.net/git/gitweb.cgi?p=cdk
10:48 zarah sneumann's link is also http://tinyurl.com/dnyxao
10:58 maclean I suppose - but that is the record of commits, not the repository.
10:59 maclean well, I'm not sure, really. The sourceforge git docs don't mention an http url.
10:59 sneumann sic. I found http://www.gelato.unsw.edu.au​/archives/git/0605/20511.html
10:59 zarah sneumann's link is also http://tinyurl.com/cf9fos
10:59 sneumann and currenty talking with net admins
11:00 sneumann So I'll solve the problem on our side.
11:00 maclean ah, great.
11:09 sneumann http://limlian.blogspot.com/2008/08/use​-http-proxy-for-git-client.html?showCom​ment=1221973140000#c6839850601346705219
11:09 zarah sneumann's link is also http://tinyurl.com/d37del
11:09 sneumann socat TCP4-LISTEN:9418,reuseaddr,fork PROXY:proxy.company.com:cdk.git.so​urceforge.net:9418,proxyport=3128 &
11:09 sneumann git clone git://localhost/gitroot/cdk cdk
11:10 sneumann but that needs squid to allow CONNECT for 9418
11:10 sneumann otherwise
11:10 sneumann 2009/04/30 13:06:22 socat[15264] E CONNECT cdk.git.sourceforge.net:9418: Forbidden
11:10 sneumann :-(
12:15 maclean joined #cdk
12:33 Conrad joined #cdk
14:12 bag joined #cdk
14:55 egonw joined #cdk
15:15 maclean joined #cdk
15:50 rajarshi joined #cdk
16:02 maclean hi rajarshi
16:03 rajarshi maclean, hi
16:03 zarah rajarshi: You have new messages. Write '/msg zarah @messages' to read them.
16:10 Conrad joined #cdk
16:20 rajarshi maclean, aah indeed, removeHydrogens is equivalent to explicit -> implcit
16:20 rajarshi a renaming is in orde
16:20 rajarshi r
16:20 maclean true
16:20 maclean or, separate functions?
16:21 maclean would a removeHydrogens method that didn't update the implicit counts be useful?
16:21 maclean it would be more like "stripHydrogens", I suppose.
16:21 rajarshi yes, but then wouldn't the H count become indeterminate? would it be possible to add hydorgens after that?
16:22 maclean but certainly a method called "convertExplicitHydrogensToImplicit" to mirror the ImplicitToExplicit would be good.
16:22 maclean I guess that the counts would be 0.
16:22 maclean And, I don't know
16:23 maclean (I don't know if it would be possible to add hydrogens after)
16:23 rajarshi i definitely support the renaming
16:23 rajarshi i suppose H additon could be derived from atom type information
16:24 maclean well, there are bits of code throughout the cdk where atom type is used
16:25 maclean if you search for the string "Explicit" (not in google images, though :) in all the java files, you can see.
16:26 rajarshi :)
16:26 rajarshi stripHydrogens certainly makes sense with a 0 for implicit H count
16:31 Conrad joined #cdk
16:39 rajarshi maclean, re the alpha pinene bug, the attached mol files seems weird
16:39 maclean well, it's a cml file, as you pointed out
16:40 rajarshi but trying to view via SMILES conversion gives a bad depiction in daylight
16:40 maclean oh, right. I added a comment to the bug report.
16:40 maclean I blame babel
16:41 rajarshi yes, i was wondering
16:41 maclean but, the point is that the Inchi works. I will change the attached file to mol.
16:42 maclean bioclipse really needs a makeMoleculeFromInchi method in the inchi manager.
16:42 rajarshi is it possible to see a smiles representation - to visually see where the difference is from alpha pinene?
16:43 Conrad joined #cdk
16:45 maclean sure. it is er...C1(C)=C(C)CC(C)C2CC21
16:46 maclean if only there was a dynamic smiles generator....
16:46 maclean (apart from the daylight one, which I just used to verify that :)
16:46 maclean your one seems not to be working at the moment
16:48 rajarshi back up
16:51 rajarshi can you guys runs the REST service jar file? redundancy would be nice
16:52 maclean hmmm. possible I suppose.
16:52 maclean heh try:
16:52 maclean c12c3c4c5c1c1c6c7c2c2c8c3c3c9c4c4c%10c5c5c1c​1c6c6c%11c7c2c2c7c8c3c3c8c9c4c4c9c%10c5c5c1c​1c6c6c%11c2c2c7c3c3c8c4c4c9c5c1c1c6c2c3c41
16:52 rajarshi as Neo said, woah
16:53 maclean looks like spaghetti :)
16:53 rajarshi ha ha
16:53 maclean I think it's some fullerene
16:53 maclean don't really see what the 2D layout can do about that...
16:54 rajarshi i've wondered about implementing something like SPE for 2D coord generation - supposedly handles stuff like fullerenes etc pretty easily
16:55 maclean spe?
16:56 rajarshi stochastic proximity embedding
16:56 maclean woah
16:56 rajarshi :)
16:56 rajarshi actually really trivial algorithm
16:56 rajarshi becomes a litltle more involved when you include constraints for 2D depictions, but not too hard
16:57 maclean looks interesting
16:57 rajarshi http://cran.r-project.org/web/packages/spe/
16:57 zarah rajarshi's link is also http://tinyurl.com/cyuaa8
16:57 maclean "Because it seeks an embedding that is consistent with a set of upper and lower distance bounds, SPE can also be applied to an important class of distance geometry problems including conformational analysis, NMR structure determination, ligand docking etc. "
16:58 rajarshi yes, lots of cool papers by agrafiotis on this
16:59 maclean it looks very short in c
16:59 rajarshi yes, the basic embedding code is very trivial as the constraints are vry simple
16:59 maclean hmmm
17:05 rajarshi i wrote a test to generate smiles from your isomeric structure and the the structure take from wikipedia (both as smiles inputs) - I get different generated SMILES
17:06 maclean generate the smiles using cdk? and which structure taken from wikipedia?
17:07 rajarshi got CC1(C)C2C(C)=CCC1C2(C)C from wikipedia
17:07 maclean the smiles "C1(C)=C(C)CC(C)C2CC21" is not alpha-pinene.
17:07 maclean "CC1(C)C2C(C)=CCC1C2(C)C" IS alpha pinene.
17:07 rajarshi (removed stereochemistry)
17:08 maclean the problem is that the one that is not AP has a molecule that the cdk turns into the smiles for AP.
17:09 rajarshi yes, so using the one you just pasted as AP and the smiles you got from that CML file, I get two different generated smiles
17:11 rajarshi http://gist.github.com/104551
17:11 cpyang joined #cdk
17:12 maclean riiight. now i see what it is you are doing.
17:12 maclean hmmm.
17:13 rajarshi how is it different from what you were doing?
17:14 maclean I'm starting from the molfile, not the smiles
17:14 maclean I think the bug must be dependant on the ordering of the atoms
17:14 rajarshi hmm, can you upload a mol file
17:14 maclean so, when you read it in from a smiles, it is in an order that writes out as the correct smiles
17:15 rajarshi wasn't there a bug on so0mething related to atom ordering?
17:15 maclean not sure
17:15 rajarshi we need a atom order shuffler
17:15 maclean there is one!
17:15 maclean thanks to a certain Herr Steinbeck.
17:15 rajarshi oh! where?
17:15 maclean AtomContainerAtomPermutor
17:16 maclean in the graph package
17:16 maclean I saw it the other day, I don't just memorise all the classes :)
17:16 rajarshi :)
17:18 rajarshi have you used it? it seems to only return a single result
17:20 rajarshi aah, it's a subclass
17:20 maclean which is apparently an iterator
17:20 maclean I haven't used it, no.
17:20 maclean not yet...
17:21 rajarshi got it, it's superclass is the iterator
17:21 rajarshi aargh, I *%$^ hate IMolecule
17:23 maclean me too :(
17:23 maclean babel!!
17:23 maclean if I go cml->mol->smi with babel
17:23 maclean it fails on the mol->smi
17:23 maclean but the molfile was created by babel!
17:24 rajarshi hmm, babel seems to not like the CML then
17:25 maclean no
17:25 maclean hmmm
17:25 maclean C1(=C(CC2C(C1C2)C)C)C
17:25 maclean looks an awful lot like AP.
17:26 maclean no, it's a different one again! grrrr.
17:26 rajarshi nope, not alpha pinene
17:27 maclean daylight's layout is rubbish (at least for the online one...)
17:32 maclean no, right, it's just different from the one I drew on my whiteborad
17:33 maclean but it is definitely the one that produces the AP smiles, using the cdk alone, and in bioclipse
17:33 maclean I will upload the (fixed) molfile (actual molfile) to the bug report.
17:33 rajarshi ok
17:53 maclean got to go
17:53 maclean bye
18:01 Conrad joined #cdk
18:08 jbrefort joined #cdk
18:27 edrin joined #cdk
19:13 egonw joined #cdk
19:22 maclean joined #cdk
20:42 bag_ joined #cdk
20:50 timvdm joined #cdk
23:54 Conrad joined #cdk

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