Camelia, the Perl 6 bug

IRC log for #cdk, 2008-01-09

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

All times shown according to UTC.

Time Nick Message
00:18 dleidert joined #cdk
05:21 jbrefort joined #cdk
05:27 LaserJock joined #cdk
05:27 LaserJock left #cdk
06:58 egonw joined #cdk
07:13 rojasm moin
07:16 egonw moin
07:17 rojasm Moin egonw! I saw now you tried to cantact with me yesterday
07:17 egonw posibly...
07:17 egonw I did some preliminary work on the charges branch
07:18 egonw and introduced an interface, per your suggestion
07:18 egonw I wanted to ask about the API...
07:18 egonw I made it like this:
07:18 egonw (mom) ...
07:18 rojasm I think that it is good
07:19 egonw http://cdk.svn.sourceforge.net/viewvc/c​dk/branches/egonw/charge/src/org/opensc​ience/cdk/charges/IPartialChargeCalcula​tor.java?revision=9864&view=markup
07:19 egonw so, a partial charge calculator returns an double[]
07:19 egonw so, that it does not require to change the IAtomContainer itself...
07:19 egonw wrapper code can then assign the charges, if needed...
07:20 rojasm sounds good
07:20 egonw if you run the test suite MchargesTests
07:21 egonw you'll see that for the one class I dealt with now, there are a lot of missing tests missing
07:21 rojasm ok! I will try to make more test
07:22 egonw I also moved some parameters to the constructor...
07:22 egonw to make the API of the GsigmaChargeCalc easier to use
07:34 egonw rojasm: what remains to be done for your mf branch?
07:34 egonw do you consider it functioanlity-wise finished?
07:34 rojasm It is finish
07:34 egonw OK, good
07:35 egonw you updated against trunk/ right?
07:35 rojasm yes
07:35 egonw please make a patch between your branch and trunk/ (same revision you synched against)
07:35 egonw and post this on the CDK SF Patches tracker
07:35 egonw and send an email to the developers list, explaining what the patch does, and that you request everyone to look at it...
07:36 rojasm ok
07:36 egonw last time when I did this, I asked everyone to add comments they have to the source code, like we did too, that is, like:
07:36 egonw / FIXME: MF: this is my comment (MyName)
07:36 egonw / TODO: MF: this is my comment (MyName)
07:36 egonw fixme/todo depending on the nature of the comment
07:37 egonw these are always suggestions, and your responsibility to reply to (not necessarily comply to, but at least reply)
07:37 rojasm ok! now I remenber I have to move some files to the correct directory
07:37 egonw let me know if you needs assistance...
07:38 rojasm I will try this morming
08:29 jonalv joined #cdk
09:10 steinbeck joined #cdk
10:12 masak joined #cdk
10:26 masak joined #cdk
12:27 jbrefort joined #cdk
12:36 jpansane1 hello
12:37 egonw hi jpansane1
12:38 jpansanel We'll organize a round table for Free Software at the Summer School of ChemInfo (Obernai 2008)
12:39 masak joined #cdk
12:51 egonw do you got posters for that meeting?
12:52 jpansanel no
12:52 jpansanel I don't have
13:31 rojasm egonw around?
13:31 egonw yes
13:32 rojasm I agree MassToFormula should remain in formula package
13:33 rojasm But I put only your suggestion
13:33 rojasm But I am think it is also a tool
13:34 rojasm but ok, for me is perfect to leave where now it is
13:34 rojasm :)
13:34 rojasm Apropo I sent a mail a devel list
13:34 egonw yeah, but cdk.tools is just existing because it's a place to pile up things not big enough for a package
13:35 rojasm ok
13:35 rojasm then I agree that it should be in formula package. I change now
13:37 rojasm Ups I didn't create in build.xml the process which pack in a new module formula
13:38 jpansanel egonw : how do you generate http://rdf.openmolecules.net/?InCh=1/CH4/h1H4 ?
13:39 egonw jpansanel: PHP code
13:40 egonw jpansanel: saw the message about the chem-struct deb package
13:47 jpansanel ok
14:06 egonw rojasm: please create a patch against trunk, and upload that to the Patches tracker on sf.net/projects/cdk
14:19 rojasm egonw: I don't what you say exactly. Some think like that-> http://sourceforge.net/tracker/inde​x.php?func=detail&aid=1866829&a​mp;group_id=20024&atid=320024
14:19 rojasm sorry
14:19 egonw yes, right
14:20 egonw ummm... what was the revision of the last trunk revision you syunched with?
14:20 egonw 9869, right?
14:20 rojasm yes!
14:21 egonw ok, you can make the patch with:
14:21 egonw svn diff https://cdk.svn.sf.net/svnroot​/cdk/branches/miguelrojasch/mf https://cdk.svn.sf.net/svnroot/cdk/trunk/cdk@9869
14:21 egonw I think
14:21 egonw haven't tried the "@9869" yet
14:21 egonw but that's what the docs say :)
14:22 egonw crap..
14:22 egonw and the other way around...
14:22 egonw svn diff https://cdk.svn.sf.net/svnroot/cdk/trunk/cdk@9869 https://cdk.svn.sf.net/svnroot​/cdk/branches/miguelrojasch/mf
14:23 rojasm doing the process
14:23 rojasm :)
14:25 jonalv are the CIA commit bot in this channel out on a some secret mission?
14:26 jpansanel egonw : do you have any poster about The BlueObelisk Movement ?
14:26 egonw jpansanel: no
14:26 egonw might be an idea though
14:27 egonw jonalv: yeah, seems so
14:29 jonalv egonw, I just thought I should tell you that the biojava manager is comming along fine. We have fancy recording capability somewhat working now  so you might want to start thinking about starting something similar for cdk as you asked about earlier...
14:30 * jonalv probably broke some kind of record in long sentence construction there
14:30 jonalv or if not record so perhaps some rule... :)
14:30 masak no, it's fine
14:30 masak could have expressed yourself shorter, though
14:30 masak :)
14:32 egonw jonalv: yeah, might be interesting indeed...
14:32 egonw so, can you briefly explain the architecture...
14:32 egonw what wrapping needs to be done, and how is that done?
14:32 egonw where and how does the recording come in?
14:33 jonalv do you have half an eternity for it right now? :)
14:34 egonw yeah, I'll twist the space-time continuum a bit so that it fit in between now and 16:15 :)
14:34 jonalv okey
14:34 jonalv where to start?....
14:35 masak I'm listening too
14:35 jonalv You have had a look at the biojava.business plugin?
14:35 jonalv it declares business and domain stuff
14:35 masak hm, actually I didn't install that again in my Big Reinstall of bc2
14:35 masak should I?
14:36 masak oh, sorry
14:36 masak here it is :)
14:36 jonalv it contains a business package and a domain package
14:36 jonalv the domain objkects are aomething that I would like to talk about some other time
14:37 egonw no, did not look at that yet... but will do that in about -5 minutes
14:37 masak that's fast
14:37 egonw (remember the space-time-continuum mess up)
14:37 masak ah
14:37 * jonalv remembers himself
14:37 masak convenient
14:37 jonalv hm that didn't work great on english
14:37 masak reality hacking
14:38 masak jonalv: go on
14:38 egonw yeah, to fit all that in about 40 minutes, you have to be behind-schedule a bit... or ahead-of-time... depending on your perspective
14:38 * egonw tends to remember himself too :)
14:38 jonalv So in the business package we have the Biojavamangaer class
14:38 egonw happens every morning...
14:38 masak ...you think
14:38 egonw "oh god, who am I? oh yeah, it's Egon"
14:39 masak :)
14:39 jonalv which is the object containing all of our fancy specially tailored methods
14:39 Annzi joined #cdk
14:39 jonalv hi Annzi
14:39 Annzi hi egonw
14:39 Annzi hi jonalv
14:39 * egonw opened http://bioclipse.svn.sourceforge.net/viewvc​/bioclipse/branches/bioclipse2/bioinformati​cs/plugins/net.bioclipse.biojava.business/ 2 days ago
14:39 Annzi egonw: Ola here at Anniz computer
14:40 egonw hi Anniz
14:40 masak Annzi: man, that's confusing
14:40 jonalv There is an interface as well
14:40 masak everybody's Annzi
14:40 Annzi egonw: Anniz is working on integrating MOSS into Bioclipse as descriptors
14:40 * jonalv is not Annzi any longer
14:40 egonw yes, saw that...
14:40 masak jonalv: you mean the other way around
14:40 Annzi she might have some questions, and you are my favourite chemoinformatics teacher :-)
14:40 egonw Annzi: mom... need to get back into focus again...
14:41 * egonw was still here two days ago
14:41 Annzi so if I Ola can't answer, I (Anniz) might ask you :-)
14:41 egonw yeah, no problem
14:41 Annzi egonw: sure
14:41 jonalv The OSGI framework register services under interfaces
14:41 egonw I got hands on MoSS experience
14:41 egonw I'm really happy to see you work on that
14:41 Annzi ok, ttl
14:41 masak jonalv: what does that mean?
14:41 Annzi that's great!
14:41 Annzi will ask you tons of questions then
14:42 egonw sure
14:42 jonalv For now think of what we uses of OSGI as a big container
14:42 jonalv sort of hashmap
14:42 egonw I'll try to answer a few of them :)
14:42 * egonw moves back two days in time...
14:42 jonalv you give an interface and you get a service which i registerd for that interace
14:42 masak egonw: wb
14:42 egonw jonalv: sorry... was away for a second...
14:42 egonw jonalv: so, the manager provides the wrapping ...
14:42 jonalv (there are ways to register many objects for the same interface but ignore that for now)
14:43 * egonw is confused by all this time shifting...
14:43 jonalv egonw: yes
14:43 masak jonalv: ok, ignoring
14:43 jonalv the manager contains the wrapping
14:43 egonw / TODO Auto-generated method stub
14:43 egonw where does that come from>?
14:43 egonw from the interface?
14:43 jonalv Eclipse
14:43 masak jonalv: you should remove those
14:44 masak that's why they are TODO
14:44 egonw argh...
14:44 egonw so we have another wrapping layer...
14:44 jonalv I use them as a reminder that the method is just an autogeneretad method stub and should be filled out with something else
14:44 masak jonalv: well, shouldn't it, then?
14:44 egonw ok, so, to integrate CDK, I have to just make such a manager?
14:44 masak jonalv: you mean the TODOs are still there because they are still valid?
14:44 jonalv egonw, yes that's the idea
14:45 egonw OK, so where do I get an overview of all wrapping that is available?
14:45 jonalv masak: yes the methods are empty except for the todo:s
14:45 egonw for example, to open an existing resource?
14:45 egonw or create a new resource in some project?
14:45 masak jonalv: ok
14:45 jonalv yes
14:45 jonalv These methods are what the gui should call when you click
14:46 jonalv but they are also reachable from the scripting consoles
14:47 jonalv so when you do something in Bioclipse that action simply call the corresponding method in the correesponing manager
14:48 jonalv the fancy thing is that the recorder plugin listens and records all these things and since the scripting uses the sam api a script that odes the same thing can be generated from the action history recorded by the recorder
14:48 egonw what?
14:48 egonw [2008-01-09 15:45:55] <jonalv> These methods are what the gui should call when you click
14:48 egonw ??
14:48 egonw how does that work?
14:48 egonw you mean there will be a central interface to functionality in a library?
14:48 egonw and no custom actions in BC2?
14:49 egonw actions, as in RCP actions
14:49 egonw actions, as in RCP actions code behind menu items/buttons/ etc
14:49 jonalv They will be a one line wrapping of a manager call
14:49 egonw well, surely that is needed for the recording...
14:49 egonw OK, major redesign
14:50 egonw did that pass the mailing list?
14:50 jonalv yes and both the gui and the scripting environemt uses the same code
14:50 jonalv so write once use twice
14:51 masak egonw: this is one of the central tenets of bc2
14:51 jonalv eh it might have gone the fast line
14:51 jonalv but it is on the wiki
14:51 masak I'm not so sure it'll work to a worthwhile degree, but I'm hopeful and definitely ready to try
14:52 masak if it work's it'll be so cool
14:52 jonalv yes so cool
14:52 masak I mean, wow...
14:52 * masak faints
14:52 * jonalv has his own private fan-club... :)
14:52 masak ok, so back to the harsh reality
14:53 masak today we've (mostly jonalv++) started connecting everything
14:53 masak amazingly, so far it's working
14:53 jonalv so egonw are you still here?
14:53 masak there's still a lot to do
14:53 egonw yes
14:53 egonw ok
14:53 masak but the basic proof-of-concept functionality is there
14:53 egonw right
14:53 jonalv as of today
14:54 egonw so, how does one then create menu items?
14:54 egonw if I have to start porting code...
14:54 masak rhino.shell.getMenu.add(...)
14:54 jonalv ah that's the next step
14:54 masak :)
14:54 egonw so, the whole GUI will be a script??
14:54 jonalv what is working right now is from the console
14:54 egonw wel...
14:54 egonw really not a bad idea...
14:54 egonw Jmol uses that too...
14:54 jonalv no it's more like everything you do in the gui can be transalted into a script
14:55 egonw menu items have a label and a bit of Jmol script to run when the menu is called
14:55 masak egonw: I found that something like that works
14:55 masak egonw: unfortunately closing the shell works too... :(
14:55 egonw :)
14:55 egonw sure
14:55 jonalv and not just any script but since it's just an api it could be into js, ruby or groovy
14:55 * egonw coughs *beanshel*
14:55 jonalv yes
14:55 masak getting to that
14:55 masak Real Soon Now
14:55 egonw excellent...
14:56 egonw :)
14:56 jonalv that seem to have been forgotten somewhere but it should be possible yes
14:56 egonw so, if I go back to the now, it'l be finished?
14:56 egonw remeber it's only Monday :)
14:56 egonw well, here at least...
14:56 egonw this time warping is always confusing
14:56 masak egonw: many times over
14:57 jonalv s othere is a lot of "magic" spring AOP and osgi stuff going on in order to get this to work easily
14:57 masak I think I will still be confused in two days when I reach today
14:58 jonalv The other thing that I ahven't done (or told you about yet) is that we are going to need our own domain objects
14:58 jonalv we need an unique id on every object we are working with
14:58 egonw what's a domain object?
14:58 jonalv so we need a base class
14:58 jonalv something like BioObject
14:59 masak egonw: a domain object is simply a descendent of BioObject
14:59 masak from the point of view of the model, that is
14:59 jonalv I am currently writing the interface IBIOObject and will need some feedback on that later
14:59 egonw ah, ic
14:59 jonalv domain objects are the objects containing the data
14:59 egonw so a MoleculeObject
14:59 egonw SpectrumObject
15:00 egonw SequenceObject
15:00 egonw etc?
15:00 jonalv managers are the objects containing the mthods,
15:00 jonalv it's like mvc
15:00 jonalv model = domain
15:00 jonalv controller = managers
15:00 jonalv view = gui or console
15:00 egonw very good
15:00 jonalv exaclte
15:00 jonalv exactly
15:01 jonalv So the molecule will probably wrap a cdk atomcontainer or something like that
15:01 egonw right
15:01 egonw excellent
15:01 egonw this nicely matches the discussion on editor we had earlier
15:01 jonalv it does? great :)
15:02 jonalv I think this is a sound working model
15:02 egonw ok, let's close this session for now...
15:02 egonw will experiment with it later this month...
15:02 egonw when all the reporting and grant writing has finished
15:02 jonalv I am a bit worried about how much will be done in editors and how much will be done thgourgh other gui and how persisting and such will work but I think it should be solvable
15:03 jonalv since we have unique id's on everything
15:03 jonalv we should be able to keep things apart
15:03 egonw yeah, the editors should be using this API too...
15:03 egonw but I garantee this will not work for a lot of editors...
15:04 jonalv yes, exactly...
15:04 egonw JCP has way too many edit options...
15:04 egonw we are slowly moving that way...
15:04 egonw but, *really* a lot of work
15:04 jonalv of course :)
15:04 egonw anyway, got to close this space-time warp now
15:04 jonalv oki
15:04 egonw let's meet in two days?
15:04 * egonw warps back to the now
15:05 egonw jonalv: that was a really interesting chat last Monday :)
15:05 jonalv =)
15:05 jonalv so what do you think now that you have had some time to think it over? ;)
15:06 egonw yes
15:07 egonw all I now need to know is how to register the manager and the domain objects
15:07 egonw each plugin will need to be able to do that
15:08 jonalv the manager is registered as an osgi service through spring-dm and as for the domain objects I am not quite sure how much registering is needed
15:09 egonw how does that work?
15:09 egonw how will the script interpreter know which domain objects exist?
15:09 egonw or does that not matter?
15:09 jonalv when would it matter?
15:09 jonalv and how?
15:10 jonalv you will need a reference to the object in order to use it...
15:10 jonalv and if you have a reference you know it exists
15:11 jonalv The scripting environment is using edrin++'s static import of classes if that is what you mean
15:11 jonalv so it can use all domain types that are in our plugins
15:11 jonalv right masak?
15:12 * masak is thrown back into the converation
15:12 masak um, well...
15:12 masak yes! right.
15:15 egonw but anyone can add domain objects via new plugins, right?
15:15 jonalv yes that's the idea
15:16 jonalv but I don't see any use of an extension point
15:16 jonalv (as of today)
15:16 jonalv there is an extension point for manager objects though
15:17 jonalv you can make manager objects that works both in the scripting and in the gui without spring except the recording will of course not work so that should only be for temporary uses
15:18 jonalv (tht extension point is used for getting the manager objects into the scripting environments)
15:25 * egonw notes that the cia.vc site is down too
15:42 egonw rojasm: did you upload the patch yet?
15:42 egonw by any chance?
15:43 rojasm Not yet, because I am trying to make that all test are working for module formula and I found that in getHTMLString doesn't work
15:44 rojasm also I try to repair ;(
15:44 egonw ah, good
16:11 rguha joined #cdk
16:29 rojasm egonw around?
16:30 rojasm Could you make a look in MformulaTests. The test FormulaCoverageTest is the only which fails. I don't find the reason
16:59 rojasm left #cdk
17:19 Annzi left #cdk
18:24 carsten_ joined #cdk
18:28 dleidert joined #cdk
18:44 egonw_ joined #cdk
19:00 egonw_ bbl
21:24 carsten joined #cdk
21:27 edrin joined #cdk

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