(8:00:37 PM) Avinash: Good morning / afternoon / evening everyone. (8:00:45 PM) jrseti: Hello. (8:00:49 PM) afeder: hello (8:01:07 PM) sigblips: Hi. (8:01:17 PM) jrseti: I was tasked with providing a short software status and a ToDoList. Have a look at http://174.129.14.98/os_plans/ (8:01:21 PM) Avinash: I have to leave a few minutes before noon. So, if we are not done, I will quietly drop out. (8:01:32 PM) jrseti: We can discuss (8:01:34 PM) afeder: thanks jrseti (8:01:34 PM) Avinash: Thanks jrseti. (8:01:40 PM) Avinash: Let us collect agenda items. (8:01:51 PM) Avinash: 1) jrseti's status and ToDo list. (8:01:59 PM) Avinash: 2) ? (8:02:20 PM) jrseti: I also may have to bug out early de to a doc appt made 6 months ago (8:02:28 PM) afeder: any news on the AI's from last week? (8:02:31 PM) MichaelM: sigblips has a useful suggestion on sonata code modularization (8:03:00 PM) Avinash: 2) SonATA code modularization, 3) ? (8:03:16 PM) Avinash: 3) AI's from last week, 4) ? (8:04:01 PM) Avinash: OK. jrseti - the "floor' is yours. (8:04:22 PM) afeder: 4) if there's time i have a little idea regarding AWS i'd like to discuss (8:04:47 PM) jrseti: Can everyone look at http://174.129.14.98/os_plans/ (8:05:37 PM) jrseti: This was the best status I could think of presenting. basically we are doing final testing and fixing small things we are finding. (8:06:13 PM) afeder: looks good (8:06:30 PM) jrseti: The ToDo list may be more interesting to discuss. This was just a start of a list, and I expect more ideas to come from you guys (8:07:00 PM) Avinash: jrseti: is the todo list available on line for people to add to it? (8:07:02 PM) Jill entered the room. (8:07:10 PM) jrseti: I can take ideas or comments here or by email and create a better ToDo list next week (8:07:29 PM) jrseti: I purposely did not publish the URL anywhere yet (8:08:04 PM) jrseti: Is there a private place to post this ToDo list? I can create a small WiKi we can all access, if you like (8:08:56 PM) Avinash: Is there any issues with posting it at a public place? (8:09:10 PM) afeder: the URL is going to be posted when (if?) I post the chat log from this meeting (8:09:45 PM) jrseti: Do we want to place this on our main WiKi? (8:09:58 PM) afeder: sounds fine with me (8:10:30 PM) jrseti: OK, I can do that (8:10:44 PM) jrseti: In the Wiki? Of in the forum? (8:11:07 PM) afeder: whichever you prefer (8:11:09 PM) Avinash: We should also distinguish between those todo's that can be done now, vs. those that have to wait until the software is open sourced. (8:11:56 PM) Avinash: Wiki may be better as there will be one place with the latest state, whereas on the forum, you will have to read multiple mails to get a complete picture. (8:12:03 PM) Avinash: What do you think? (8:12:31 PM) jrseti: I assume a WiKi post AND a forum topic. So anyone can comment, and Wiki editors can add to the Wiki (8:12:38 PM) afeder: +1 (8:13:00 PM) jrseti: OK, I'll do that (8:13:31 PM) Avinash: Good. Any comments on the content of jrseti's list? (8:13:37 PM) sigblips: Re: Telescope and SonATA status stream. Human readable text would be good. A couple lines of code could post this to Twitter and an IRC channel. Appending status changes to a text file in the cloud is good too. (8:13:55 PM) Jill: who decides who works on which of the 'todos'? (8:14:18 PM) MichaelM: Might be nice to have target dates. (8:14:57 PM) Avinash: Right now, let us keep it informal. So, if someone offers to work on something, they put their name on the Wiki, and start a discussion about it. (8:15:19 PM) Avinash: Target dates are good, especially for items that others depend on. (8:15:31 PM) jrseti: I think OS contributors will have to give us a short proposal telling us how they want to accomplish an item, and a proposed finish date. But this list is informal now, and will only kick in when we release the source code. (8:16:42 PM) Avinash: Is there anything they can do now - before we release the code? (8:17:33 PM) jrseti: Not really, we have to manage this and we don't have time at the moment. By "we" I mean the software team (8:18:52 PM) Avinash: OK. (8:19:26 PM) jrseti: So, I'll post this week and we can continue the ToDo list on-line (8:20:57 PM) Avinash: Sigblips - do you want to talk about your proposal for SonATA code modularization? (8:21:25 PM) MichaelM: eric, is it okay that I mentioned it? (8:22:08 PM) sigblips: The idea is to split SonATA up into smaller parts that might be of value to someone. Such as the channelizer and beamformer code. (8:23:13 PM) sigblips: My concern is that the SonATA source is going to be of little value to anyone who doesn't have a radio telescope and a grid of computers. (8:23:30 PM) MichaelM: What I liked about the scheme is that it creates discrete chunks of value that folks might invest in maintaining (8:23:55 PM) jrseti: The chaneelizer and dx programs are separate entities (8:24:13 PM) MichaelM: great! then we could proceed with that plan directly! (8:24:33 PM) Avinash: MichaelM - but greater modularization helps, so we could do more. (8:24:49 PM) jrseti: Basically i see it as already modularozed into seeker, dx, channelizer, packetgen, packetrelay, etc. (8:24:51 PM) Avinash: My suggestion is that we os it, the way it is, and then seek advice from the community on what parts to modularize more. (8:25:50 PM) sigblips: Turn the channelizer into a distinct library and someone could take it and plug it in easily to their project. This could create broader adoption and some praise for the SETI Institute. (8:26:34 PM) MichaelM: sometimes leadership is needed. To quote Henry FOrd "If I'd asked my customers what they wanted, I'd have just tried to sell a faster horse." (8:27:05 PM) jrseti: Yes, I assume we will want to create distinct libraries like that, Sigblips. (8:27:24 PM) sigblips: Sounds good. (8:27:35 PM) Jill: a black colored horse! (8:28:01 PM) jrseti: That may warrant a plan, or ToDo list, for the OS community to do this type of modularization. (8:28:04 PM) MichaelM: good blending of quotes, Jill! (8:28:30 PM) jrseti: Sigblips, we can add this type of issue to the ToDo list once I get it on the WiKi (8:28:38 PM) sigblips: The choice of license will also have a big effect on how these libraries are adopted. GPL or LGPL might be good depending on goals. (8:28:56 PM) MichaelM: One nice aspect about going out with the beamformer or channelizer standalone is it gives us a chance to test the waters. (8:29:45 PM) MichaelM: ooh, another ford quote: Nothing is particularly hard if you divide it into small jobs (8:29:52 PM) ***MichaelM shuts up now (8:30:25 PM) sigblips: Just keep dividing by 2 and everything becomes trivial. (: (8:30:38 PM) jrseti: I'll make a separate section in the ToDo list for modularization (8:31:08 PM) sigblips: That's a O(n log n) joke if you didn't get it. (: (8:31:59 PM) Jill: we keep talking about the signal processing chunks like channelizer - but it's the seeker that integrates all the info about candidates and RFI and makes a 'detection' decision. we are now talking about giving the community more tools and cranks to turn, but not really the smarts to interpret results (8:33:06 PM) MichaelM: that is true. But some of our coding contributors are looking for a screwdriver, not a swingset kit. (8:33:30 PM) jrseti: If we package up the channelizer as a "product", we could potentially monetize on this for non-SETI related products. (8:34:36 PM) sigblips: I don't know much about the "seeker" but it sounds complicated. Is it possible to split it up into smaller consumable re-usable chunks? (8:35:41 PM) Jill: i understand, but we will encourage contributors into publishing false positives. i think modules are great for use in other pursuits. for SETI i hope we can help community to be thoughtful consumers (8:37:12 PM) jrseti: It may be possible to improve the code through more modularization. But that is a big issue and will require a community of people to tackle the issue. (8:37:23 PM) sigblips: If monetization is the goal then GPL is a good choice. If broad adoption and possible backward improvements / bug fixes is the goal then LGPL is a better choice. (8:37:55 PM) jrseti: Any progress on the licensing issue? (8:38:38 PM) MichaelM: had a nice chat with eric yesterday (8:38:49 PM) Avinash: Eric? (8:38:54 PM) MichaelM: Sigblips (8:38:56 PM) MichaelM: sorry (8:39:14 PM) Avinash: Outcome of the talk? (8:39:24 PM) sigblips: Jill what do you mean by "encourage contributors into publishing false positives?" (8:39:38 PM) MichaelM: one challenge, a major one, is that it's not clear at all how SETI wants to "monetize" SonATA (8:40:01 PM) MichaelM: this affects some of the license options (8:40:23 PM) Avinash: MichaelM: That is correct. All we are trying to do right now, is to not take a path that will exclude future monetization. (8:40:37 PM) MichaelM: Okay, no open source. Next? (8:40:41 PM) Jill: sigblips - there will be LOTS of signals detected, most or maybe all will be RFI, much of it previously known. without the tools these signals will be published as candidate ETI signals. (8:41:09 PM) MichaelM: no, the real issue is that we can guess, and pick a scheme (8:41:14 PM) sigblips: Jill: OK, thanks I understand now. (8:41:21 PM) Avinash: MichaelM: OS and monetization are not incosistent. RedHat has proven it. (8:41:42 PM) MichaelM: I am aware, that was humor (8:41:50 PM) Avinash: I know. (8:41:54 PM) sigblips: Red Hat's revenue model works because they are at the end of the funnel. (8:42:30 PM) MichaelM: If SETI wants to make money off control of the source, OS is a poor choice. However, if we want to make money off delivering a custom job, then some OS may work (8:42:42 PM) Avinash: Back to the topic. On the license issue, all I can say is that there has been a lot of discussion, but no resolution yet. Do you agreed MichaelM and Afeder? (8:42:52 PM) MichaelM: ? (8:43:24 PM) afeder: yes, no resolution yet - ball's in your court though (8:43:32 PM) Avinash: Agreed. (8:43:54 PM) jrseti: All who want GPL, raise your hand (8:44:00 PM) sigblips: My only concern is choosing GPL because of a conflict with FFTW is a mistake. That said GPL might be the best choice for other reasons and then it might not be. (8:44:16 PM) MichaelM: i agree with sig (8:44:25 PM) Avinash: FFTW is just one factor. And, we would not choose a license just for FFTW. (8:44:41 PM) Avinash: You all know that we had started looking at alternatives to FFTW. (8:45:42 PM) MichaelM: Seems to me that the only rational choices are GPL, LGPL, APache-style, and "just take it". Each of these can be done in a dual-license scheme, though the latter two would be stupid. (8:45:47 PM) afeder: jrseti: i'd prefer GPL, but not at all costs .. viability of the project is first priority (8:46:38 PM) MichaelM: +1 (8:46:41 PM) sigblips: I think it all comes down to what are the SETI Institutes goals? I think the goal is detecting SETI and everything else is secondary but I could be wrong. (8:47:19 PM) Avinash: Sigblips - agreed. (8:48:35 PM) sigblips: Dual license is a good option. (8:49:03 PM) Jill: but we need to keep searching until we do detect a signal, so monetizing needs to raise its ugly head somewhere (8:49:58 PM) jrseti: I have to bug out - a doc appt made 6 months ago... (8:50:09 PM) afeder: see you jrseti (8:50:11 PM) afeder: what is the argument against GPL, again? (8:50:20 PM) afeder: dual-license GPL (8:50:33 PM) jrseti left the room (quit: Quit: Page closed). (8:51:01 PM) afeder: if there is no significant argument against it, maybe just roll with it? (8:51:42 PM) MichaelM: It's sort of a poison pill for some, but they can come back to us for a license (8:52:05 PM) Avinash: Just like FFTW. (8:52:06 PM) sigblips: Dual license GPL would require every contributor to sign over copyright much like what is done with GNU projects. Some people might not like doing that. (8:52:14 PM) afeder: but they don't have to consume the GPL code in the first place (8:52:43 PM) afeder: sigblips: only if SETI wants to resell contributions (8:52:51 PM) MichaelM: I do not agree, sigblips. A non-exclusive license will do us just fine (8:52:52 PM) Avinash: No, they don't have to sign over copyright - just give us a non-exclusive perpetual licence. (8:53:31 PM) Avinash: OK five more minutes for me. (8:53:43 PM) sigblips: Anders: True but I think monetization is the goal for going the dual license route. (8:54:02 PM) Jill: has a decision finally been taken? (8:54:18 PM) MichaelM: we can resell for $$$ based on a nonexclusive license of contributions (8:54:42 PM) Avinash: I like the $$$ vs a plain old $. (8:54:43 PM) afeder: MichaelM: what Avinash proposing is selling only SI-developed code (8:54:50 PM) afeder: Avinash: lol (8:55:17 PM) MichaelM: Yes, I think that's not realistic. Contributed code finds its way in there. Document it right (8:55:28 PM) Avinash: These are alternatives we will have to look at. (8:55:47 PM) afeder: "finds it way in there"? really? (8:56:01 PM) MichaelM: really. Coders are cannibals (8:56:10 PM) Avinash: Yes, look at the open source movie "Sita Sings the Blues." (8:56:35 PM) Avinash: In order to distribute the movie, you have to get to get license from lots of people. (8:56:57 PM) afeder: well, that's not a good analogy to selling just SI-developed parts (8:57:13 PM) Avinash: OK folks, I am out of time, and have to drop off. I will read the transcript later. (8:57:21 PM) MichaelM: ok (8:57:25 PM) afeder: alright see you (8:58:12 PM) afeder: MichaelM: FFTW does this - selling just their own code and releasing it GPL at the same time (8:58:18 PM) MichaelM: it's all well and good to plan to just sell SI-developed, and maybe we can do that. However, why on earth not seek to get adequate licenses back so that we are not shut down (or forced to split the REVENUES 50/50) if our team errs? (8:58:19 PM) sigblips: "Sita Sings the Blues" is available on Netflix. I guess they worked out the licensing issues! (: (8:59:27 PM) MichaelM: interesting (8:59:32 PM) Jill: and our team will err --- as you say, coders are cannibals, without realising it. (8:59:52 PM) MichaelM: afeder: i know! (9:00:15 PM) afeder: Jill: so are FFTW superhuman non-cannibal coders? or how do they manage to pull it off? (9:01:09 PM) Jill: AFeder - don't know - i just know we were surprised when we actually looked at our own code from that point of view (9:01:12 PM) sigblips: MIT's FFTW is a good revenue model to emulate but the problem is that nothing in SonATA is that generic, it is all extremely niche. My guess if MIT's makes a lot less than $100K / year on FFTW. (9:01:27 PM) MichaelM: normal software development practices. you have a team, you quiz them, you look at the code, and you release under whatever license you pick. (9:01:55 PM) MichaelM: the trouble comes when you start blending code from multiple sources. Things ... migrate. (9:02:30 PM) Jill: i need to migrate down HWY 880 now - bye. (9:02:37 PM) afeder: see you Jill (9:02:40 PM) MichaelM: hv fun (9:03:05 PM) Jill left the room (quit: Quit: Page closed). (9:03:06 PM) afeder: okay, how about instead of CLA then BSD or other permissive license? this would allow SETI to cannibalize (9:03:46 PM) afeder: permissive licensing for contributions that is (9:03:52 PM) MichaelM: might be the trick. By CLA, do you mean the apache license? (9:04:04 PM) MichaelM: I can take a closer look at BSD for this (9:04:20 PM) afeder: by CLA i mean a custom-written agreement that gives rights only to SETI (9:04:31 PM) afeder: that has to be signed and all (9:05:07 PM) MichaelM: ah yes. No, I think we can go with something less demanding (9:05:20 PM) MichaelM: and more open (9:05:33 PM) afeder: that would help a lot, in my opinion (9:05:59 PM) MichaelM: yeah, that "assign copyrights" thing is pretty wrong in my view (9:06:07 PM) afeder: yes (9:06:31 PM) MichaelM: Signed is nice, but we can tread an easy middle ground with electronic signatures. (9:06:43 PM) sigblips: Remember that the GPL is not actually a license but it is copyright. (9:07:05 PM) MichaelM: Uh, i don't think that word means what you think it means. (9:07:24 PM) sigblips: inconceivable (9:07:30 PM) MichaelM: heh (9:07:52 PM) MichaelM: GPL is a License to use a copyrighted work under specified conditions (9:08:14 PM) MichaelM: (use) = copy, publish, perform, sell, distribute, etc (9:08:28 PM) sigblips: They just say that to make you think it is a license. (: (9:08:56 PM) MichaelM: \me stamps foot. "It IS, it is it is!" (9:09:15 PM) MichaelM: isn't it? (9:10:08 PM) sigblips: It's not a contract license is what I meant but these are legal things I know nothing about. (9:10:28 PM) MichaelM: I know a lot about them. It's a license and also is a contract. (9:10:58 PM) sigblips: Well you are the expert so I'll take your word on it then. (9:11:38 PM) MichaelM: yay! for my next trick i shall explain how a FFT works! (9:12:03 PM) sigblips: We already went over that. It's easy just keep dividing by 2. (9:12:54 PM) MichaelM: heh. it's all ones and zeroes (9:14:06 PM) MichaelM: we got aught else? (9:14:52 PM) afeder: whats that? aught=anything? (9:15:30 PM) sigblips: "anything at all" I had to look it up. (9:15:45 PM) afeder: ah :) (9:16:07 PM) MichaelM: oop (9:16:08 PM) sigblips: It olde English. (9:16:13 PM) MichaelM: aye (9:16:23 PM) MichaelM: sorry, feeling piratical here (9:16:25 PM) MichaelM: mateys (9:16:38 PM) afeder: oi oi (9:16:56 PM) afeder: nope, i guess thats about it (9:17:44 PM) MichaelM: all righty then. i will look at clas that don't suck (9:18:10 PM) afeder: neat (9:18:19 PM) afeder: see you next time