MODEST meeting 2009-10-07
From Micasim
| Project | SL |
|---|---|
| Project Lead | Rob Knop |
| Project Homepage | Virtual World Workshops |
Contents |
Summary
AMUSE Meeting Summaries
Piet points to Rob's summaries at [Amuse Workshop Oct 2009]. There is some discussion as to how the slides got transcribed. In fact, I (Rob) just type pretty fast.
Simon (Lakhesis Destiny) dropped by to report that things are going well at the AMUSE workshop. The hope is that all of the MUSE community will migrate to AMUSE. AMUSE is still in early stages. He says there is a sort of primer, although really it's more of a scope document. (No URL at the moment.) The newest things in AMUSE are an MPI interface (instead of f2py and swig from MUSE), a unit conversion interface, and unit testing.
There are good interfaces for N-body code now. Evert and Tassos are working on the Stellar Evolution modules. Garrelt and others are working on the Radiative Transfer and Gas Dynamics interface.
AMUSE Licensing
There was discussion about licenses. (The summary includes some of what was discussed in-world as well as what Rob remembers from the meeting.) There are a few conflicting needs.
- AMUSE wants to be absolutely as open source as possible.
- Many of the invidual package authors (especially younger folks) need to protect their work so that they get credit (citations or coauthorship as appropriate) if it is used by other scientists.
Some of the software codes included ("legacy codes" in the parlance of AMUSE) aren't going to be so freely licensed. In particular, some folks are going to want to say that anybody publishing results of using these codes should contact them, and perhaps even offer co-authorship. We considered that this sort of thing is covered under scientific ethics anyway.
There are lots of complications from whether or not including the legacy codes is "simple aggregation" or more like "linking". If it's simple aggregation, we could GPL (say) the AMUSE code proper, and then let each legacy code be under whatever license it is. In the end, the decision was that we need to be as clear as possible who needs to be contacted or cited when various modules are run. Ideally, each legacy code's interface will have with it a very simple (and maybe automatically run) way of spitting out the credits for that legacy code. The AMUSE code itself will then probably be GPL.
Note that there are some misconceptions in the chat log. Piet says one purpose of licensing is "to protect the software from being grabbed and then possibly licensed by someone else (which I think is the main point of the GNU and MIT, etc. open source kind license)". In fact, that is the primary difference between GPL and MIT-type licenses. The GPL license says that if anybody publishes modified binaries, they also have to publish the modified source code under the GPL license. Any derived works of a GPL code must be released back to the community under the same terms it was acquired. MIT and BSD licenses, on the other hand, allow free use and redistribution of the code, but allow proprietary derived works to be used and sold without releasing the source code changes. As such, code that is BSD or MIT licensed can be grabbed and licensed by somebody else. Both kinds of code may be sold commercially. The GPL doesn't prevent sale, sale of support, of the code itself or of modified versions of the code.
Chat Transcript
Pema Pera: I'm just looking at the link that Rob sent us yesterday, http://www.rknop.net/modest/index.php/AMUSE_Workshop_Oct_2009
Pema Pera: there is a *huge* amount of information there!!
Dali Waverider: tres amusant
Pema Pera: Rob must have been extremely active, writing this all down while the meeting was going on ! ! !
Pema Pera: or did others write part of it? I see code chunks and such that would be hard to copy down from a slide
Dali Waverider: perhaps there were conference materials available in digital form
Lagrange Euler: That's a remarkable amount of data. And the level of organization is very un-MUSE like.
Pema Pera: or he used a digital camera?
Lagrange Euler: Must be Inti's doing.
Dali Waverider: and OCR. B-)
Pema Pera: Hi Lakhesis!
Desdemona Enfield: Well, whatever, it is good rapportage.
Lagrange Euler: Speak of the devil...
Pema Pera: yes, it's wonderful!
Lakhesis Destiny: Hi Piet
Pema Pera: Simon, we are admiring the very detailed report of the conference
Pema Pera: http://www.rknop.net/modest/index.php/AMUSE_Workshop_Oct_2009
Pema Pera: did you see it, and do you know whether that is all Rob's writing?
Pema Pera: Simon?
Lakhesis Destiny: Ja
Lagrange Euler: Hi Simon!
Lakhesis Destiny: Hi SteveOh. Yes, the workshop is great. Regretfully, Rob's computer died today.
Lakhesis Destiny: I think that we make good progress. Evert and Tassos are working together, and we had many good discussions.
Lagrange Euler: Good. I'm glad they could both be there.
Pema Pera: yes indeed!
Lakhesis Destiny: Oh yes, and Even Garrelt dropped by
Pema Pera: glad to hear that
Lagrange Euler: Who is Even Garrelt?
Lakhesis Destiny: Yes, I think rob os doing a great job.
Pema Pera: did you see his writeup, Simon?
Lakhesis Destiny: Not really yet.
Lagrange Euler: How many modules are now updated to use the new interface?
Pema Pera: (and yes, who is Even Garrelt?)
Lakhesis Destiny: Garrelt Mellema
Pema Pera: hahaha, okay, capitalization problem
Lakhesis Destiny: He is a Hydro and RT person from Helsinki, used to be a student of Vincent Icke. Good guy.
Pema Pera: as for the writeup, that could easily be converted into a prototype AMUSE manual/primer
Pema Pera: if not an actual one already
Pema Pera: I bet you don't have one yet?
Lakhesis Destiny: We have an AMUSE primer ready
Pema Pera: really? Great! What is the URL?
Lakhesis Destiny: But at the moment the primer is mostly a scope document. We just started with AMUSe, and a LOT still has to be done.
Lakhesis Destiny: We have been working on the unit conversion interface, the unit testing and the basic MPI interfaces, as suggested by Steve.
Pema Pera: the nice thing about the write up that Rob sent us a link to is that it is already in the form of a story, quite readable
Lakhesis Destiny: I hope that Ever tand Tassos and Arjen can come up with a unified SE interface.
Lakhesis Destiny: WEll, I browesed throught the website, and I wil lread it in more detail. It's good that it is online.
Pema Pera: Do you want to give some more impressions, Simon, about what was especially new and noteworthy, of all the developments?
Lakhesis Destiny: Well, this is really not the rightmoment to concentrate on any presentation about this. I expect that Arjen will attend regular SL meetings to give future opdates of our programming efforts, and otherwise Inti will be attending.We work very hard on AMUSE, and we make good progress, but also, such progress is needed ans the funding will run out in two years.
Lakhesis Destiny: At this moment we hope that he entire MUSE community will move to use the AMUSE framework, despite the slight differences in the interface and structure.
Lakhesis Destiny: The professionals working on AMUSE really help the development.
Lagrange Euler: How much progress has been made on bringing old modules into the new format?
Pema Pera: yes, I think it would be far better to have everybody work on a single framework, rather than having MUSE and AMUSE grow in parallel
Lagrange Euler: And how easy is it to debug the parallel interface?
Lagrange Euler: I don't think they will grow in parallel. The science modules will be the same, and the new interface whould be the preferred one, for a number of technical and practical reasons.
Lakhesis Destiny: Well, as I siad: Evert and Tassos are working with Evert on the SE modules. Garrelt and several others are working on the RT and Gas dynamics, and we had discussions about the licences.
Lakhesis Destiny: We have a few of the dynamics modules incorporated, these are mainly the old MUSE modules which we copied.
Lagrange Euler: Licences?
Lakhesis Destiny: Well, several of the young students wanted to disucss licences. I do not like the thought alone of it, but alas. I think that they will go for something like GNU or MIT licencing.
Lagrange Euler: The big hurdle will be the live SE modules. Evert seems to be on board, and I know Tassos has been thinking about parallel implementations of EZ.
Desdemona Enfield: The point of the license is toprevent a corporation from claiming the software as proprietary because they modified it.
Lakhesis Destiny: Yes, en with the new MPI moduleair interface, I think that parallel SE will be no problem. Derek is working on te parallelization and gridtification of the SE, as soon as he is finished with the CosmoGrid parallelization (we can now run on 4 computers in parallel!).
Lagrange Euler: If it is the price of having Evert and Tassos contribute, I guess a licence is OK.
Pema Pera: yes, I think you need to state some form of licence, even if only a minimal one like the MIT one (which we use for ACS)
Lagrange Euler: In our case, Des, the licence is to protect the work of recent OhDs who would ke to exercise some control over their professional life's work
Lakhesis Destiny: I do not care if a company claims the software and start making money with it. How much moeay can you make with N-bodies anyway.. As long as I can do my research I am happy.
Lagrange Euler: PhDs (!)
Lakhesis Destiny: Another matter is the contribution and citation to young people, who may need AMUSE as a lift for publications and therefore for career tracks.
Desdemona Enfield: Generally speaking these licenses require people who modify changes to the system.. add features.. to openly publish their source changes.
Lakhesis Destiny: Yes, possibly.
Lagrange Euler: The agreement is something like, "you can have the source, so long as you let me know and acknowledge me."
Lagrange Euler: I don't know if that is enough to make Evert, Tassos, and others feel secire.
Lakhesis Destiny: Indeed, Steve. That should be the intention. And cite the relevant papers. I can imagine that each AMUSE pulication should cite some 4 papers or so to do full credti.
Pema Pera: There are really two very different issues here: 1) to protect the software from being grabbed and then possibly licenced by someone else (which I think is the main point of the GNU and MIT, etc. open source kind licences) and 2) giving the authors some form of control
Lagrange Euler: This is an issue we have been thinking about since Starlab, and it is of course much older than that.
Pema Pera: If 2) is desired, that is quite a different issue from 1)
Pema Pera: but of course both can be spelled out in the same licence
Lagrange Euler: From the point of view of the author, they want their code to be recognized and widely used, and they want credit where it is due.
Lakhesis Destiny: Give authors some control, may be more relevant than we thingk. In particular if WE are not the main develpers anymore, but others take over. I think that this sure will happen. And we do not have the 150people that FLASH has.
Desdemona Enfield: These licences are intended to be supportive of community work, they are not necessarily restrictive of application uses.
Pema Pera: btw, here is what we have done for ACS: http://www.artcompsci.org/rel/mit.html
Lakhesis Destiny: I have to leave son. Next week I cannot attend, (Dagmar has to go to the dentist), but after that I'll try to be heer weekly.
Lagrange Euler: Of course, Piet, you and Jun are established researchers, with relatively little concern that someone is going to make off with your livelihood.
Pema Pera: so we have three aspects, really: 1) guaranteeing open source nature and protecting that: 2a) making sure that the original author is acknowledged and informed of the use of the code; 2b) some kind of "control" which is quite different
Lagrange Euler: (1) and (2b) may be hard to balance
Lakhesis Destiny: Good bye
Desdemona Enfield: bye
Pema Pera: bye Simon!
Ico Telling: Bye simon
Lagrange Euler: Bye, Simon
Pema Pera: We'll need to continue this discussion, it's quite central to all that we do
Pema Pera: 2a) is surely central for the career of the writier
Pema Pera: but 2b) is something quite different
Pema Pera: and if stated too strongly, probably will her the career of the writer!
Pema Pera: we have seen so many examples
Lagrange Euler: I agree. If you remember, with starlab, we started off with fairly restrictive terms, but finally relaxed them (probably as we became more secure)
Desdemona Enfield: consider this scenario: your groups develop a large softwar system... Microsoft takes it.. adds MS telescope features to it.. distributes it... then later... MS turns around an sues you for having copied there additions.. YOu now have to go to court and prove you did not.
Pema Pera: people like Volker Springer and others becoming famous exactly because everyone uses his code
Pema Pera: and others remaining unknown because they keep sitting on their code
Lagrange Euler: Sverre too. ut many people have their own private, modified versions of his codes.
Pema Pera: but the again it all depends on what Simon means with "control"
Desdemona Enfield: Is has asking for veto power on usage?
Pema Pera: yes, exactly, MUSE will not be able to compete with NBODY6 if it has significantly more restrictions
Pema Pera: we don't know Des, until we hear more from Simon :)
Pema Pera: Maybe this is a good place to stop today? We can speculate, but that won't help us further with the control question
Pema Pera: I'm very pleased to hear that Simon is planning to come here regularly
Pema Pera: so that we can discuss things easily once a week
Pema Pera: (as long as he stays long enough to actually come to a conclusion :-)
Pema Pera: Is there any last-minute question, commment?
Lagrange Euler: I think I actually raised the issue of control, since it has come up at MODEST meetings, so Simon isn't necessarily the person pushing for it.
Pema Pera: yes, but Simon is in charge, Steve, so any discussion will have to involve him
Lagrange Euler: Agreed
Desdemona Enfield: Pema, I would be glad to do some initial terraforming on your opensim region when you have it set up.
Pema Pera: Thanks a lot, Desdemona! I'm not sure when we'll set something up; probably Rob (Prospero) will be the person in charge
Pema Pera: so if you contact him, I'm sure he'll be more than pleased to hear that
Desdemona Enfield: okie.. I will send him a message.
Pema Pera: thank you!
Pema Pera: thank you all for joining us!
Pema Pera: and see you next week
