(8:01:24 PM) avinash: Good morning / afternoon / evening everyone. (8:01:39 PM) sigblips: Hello. (8:01:51 PM) jrseti: hello! (8:02:00 PM) avinash: What do we want to discuss today? (8:02:10 PM) jrseti: palamida! (8:02:21 PM) janebird entered the room. (8:02:45 PM) avinash: Sounds good. Item 1 - legal issues with open soureing the software. Item 2 - ? (8:03:04 PM) afeder: board/council (8:03:08 PM) afeder: charter etc (8:03:30 PM) sigblips: Gliese 581 (8:03:34 PM) MichaelM: heh (8:04:08 PM) parutron left the room (quit: Ping timeout: 255 seconds). (8:04:34 PM) avinash: OK 4 items. Let us start with Palamida (8:05:02 PM) avinash: Jon: Do you want to start the discussion? (8:05:16 PM) jrseti: ok... (8:05:26 PM) afeder: ("board/council" and "charter etc" is one item) (8:05:54 PM) jrseti: This last week Avinash and I have been spending a lot of time whittling away at the Palamida issues. (8:06:03 PM) afeder: awesome (8:06:18 PM) jrseti: We have about 20 left to go. We did find one BAD thing... (8:06:28 PM) jrseti: Sigblips found it... (8:06:29 PM) ***Up2 lurks, somewhat distracted by work (apologies) (8:07:22 PM) jrseti: The FFT library we use to build the programs in the binary release is GPL. So we had GPL code in our release. We pulled the release download link. (8:08:06 PM) jrseti: The Palamida report was incorrect when it listed this issue. It stated that this library was MIT licensed, but it is actually GPL. (8:08:31 PM) jrseti: What do we do? There are several options. (8:08:50 PM) jrseti: 1 - we can buy a commercial MIT license from the FFTW folks (8:10:09 PM) jrseti: 2 - We can use a substitute package called FFTSS, just for the open source version, - which is ALMOST FFTW compatible, and is LGPL. But there are a few things we would need to change in the code (8:10:10 PM) parutron_ left the room (quit: Read error: Connection reset by peer). (8:10:27 PM) jrseti: Avinash, can you say more about this? (8:10:45 PM) sigblips: I don't think they'll sell you a commercial FFTW license for open source code. I think it would break their licensing / revenue model. It is intended for closed source. (8:10:53 PM) MichaelM: Here is a question of practicality. Can we release the code under GPL while working to create and release a less restrictive LGPL and/or MIT version? (8:11:17 PM) avinash: Sigblips can elaborate more single vs. dual precision. (8:11:54 PM) parutron entered the room. (8:12:08 PM) sigblips: The FFTSS package uses 64-bit doubles. They don't use 32-bit floats which is what I suspect SonATA uses. (8:12:46 PM) jrseti: sigblips - you are correct (8:13:15 PM) jrseti: Is having someone in the community convert FFTSS to single precision a viable option? (8:13:39 PM) afeder: i does not sound like something that is überhard to do - is it though? (8:13:48 PM) afeder: it* (8:13:48 PM) sigblips: If FFTSS doesn't work for you then I'd suggest using some different FFT library. There are lots of them out there. The community might like plugging different ones in. (8:13:55 PM) MichaelM: performance hit? (8:14:25 PM) jrseti: There is a preformance hit. What - I do not know (8:14:28 PM) jrseti: Such a modification to FFTSS would probably be welcome by the FFTSS project (8:14:28 PM) avinash: The solution we figured out was to make configurable. So, if you have FFTW, you can use it, otherwise, use the default version. (8:14:33 PM) jrseti: By name FFTW is the fastest (8:15:06 PM) jrseti: We can make it configurable, as long as we can get FFTSS to do single precision (8:15:17 PM) sigblips: The problem with converting FFTSS to 32-bit floats is that it is heavy SSE3 vectorized code. So it should be possible but it won't be as simple as replacing all "doubles" with "float" variables. (8:15:38 PM) jrseti: sigblips - agreed (8:15:53 PM) afeder: i see (8:16:36 PM) afeder: as sigblips said, buying a license probably won't work either way (8:16:48 PM) jill_ entered the room. (8:16:58 PM) jrseti: Is it viable or "legal" to release our code with the FFTW3 library statically compiled - saying we want the community to fix this? (8:17:30 PM) avinash: Let us get opinion on this from a legal expert. (8:17:45 PM) afeder: GPL will "infect" the whole thing (8:17:52 PM) avinash: My view is that we don't want to get into a Cisco type siguation. (8:17:58 PM) ***afeder uses Microsoft terminology (8:18:02 PM) MichaelM: yes, but that does not forbid release (8:18:10 PM) sigblips: I think Avinash's idea of making the FFT library configurable in the Makefile is a good solution. (8:18:15 PM) jill_: Michael Fisher should be able to give opinion (8:18:16 PM) afeder: yes, GPL governs binaries too (8:19:03 PM) afeder: i agree sigblips (if interface is clean enough to do it and all) (8:19:16 PM) avinash: I would not hold much hope for releasing with GPL. For now, let us explore other options. (8:19:47 PM) jrseti: Avinash - what was the deal with purchasing a commercial license? (8:20:11 PM) avinash: We never explored it fully. We could. (8:20:47 PM) avinash: There will be some expense - could be as low as $5K, or as much as $900 per person. (8:21:17 PM) avinash: Are we saying that FFTSS option will take a lot of effort? (8:21:31 PM) jill_: does non-profit status help us? i know i dream a lot (8:21:42 PM) afeder: :) (8:21:55 PM) sigblips: You could hire SigBlips.com to 32-bit float-ize FFTSS for a lot less than that! (8:21:57 PM) avinash: Yes, it actually does. I will investigate. (8:22:13 PM) sigblips: (: (8:22:27 PM) avinash: I mean non-profit status. :-) (8:22:40 PM) jrseti: I wonder what complications this gives people who want to use this code for their own projects. Will thy have to purchase a license also? Or does our purchase clear that up? (8:23:45 PM) sigblips: That's why I'm not sure if they will sell you a FFTW license for open source code. It complicates things. (8:24:35 PM) jrseti: Avinash - could you contact them and inquire? (8:24:46 PM) avinash: Yes, I will. (8:25:06 PM) afeder: otherwise community-port and compiler option sounds like workable solution too (8:25:32 PM) avinash: Here is their non-profit license. http://web.mit.edu/tlo/www/downloads/pdf/FFTW_ver_312_Non_Profit.pdf (8:25:37 PM) jill_: any idea how long such a project might take the community? (8:26:25 PM) afeder: no, no idea - i'd have to study the problem first (8:26:35 PM) MichaelM: even the non-profit license may not work to benefit downstream users. (8:27:01 PM) jrseti: downstream users is what I worry about here (8:27:07 PM) avinash: This license seems to say that we pay only $5K, and can release the software. (8:27:40 PM) MichaelM: Yep. Actually, the users can be fine, but if they modify and distribute they have trouble, which sort of is NOT an open-source goodness (8:28:20 PM) afeder: +1 MichaelM (8:28:26 PM) sigblips: They could license it to you with a GPL exclusion like what MySQL does but that could allow other people to take it and use it in other ways thus breaking their licensing revenue model. (8:28:44 PM) avinash: Here is what it says about users: "the right to distribute, transfer, sell, lease or sublicense the right to use, reproduce or modify DERIVATIVES" (8:29:10 PM) sigblips: But does it say you can open source it? (8:29:33 PM) avinash: I think a full review of the agreement is in order. Why don't I take it as an AI, rather than trying to do it real time. (8:29:42 PM) afeder: +1 (8:29:51 PM) MichaelM: yep (8:30:03 PM) jrseti: If we wish to pursue modifying the FFTSS code for single precision, do we post this project on our site? (8:30:22 PM) afeder: jrseti: yes (8:30:31 PM) afeder: unless there's better suggestions? (8:30:32 PM) avinash: So, 2 options to be investigated: FFTSS, and MIT license. Anything else? (8:30:37 PM) MichaelM: ps, is there anything that says MIT would apply THAT license to THAT code? (8:30:41 PM) jrseti: OK - there is one more issue you may be interested in (8:31:05 PM) jrseti: Code was lifted from a book called Astronimical Algorithms (8:31:18 PM) jrseti: We are using copyrighted code from this book. I called the publisher and he said we need to get permission to release this in our open source. He will give us permission. But first I have to buy the book (out of print, I found it on the web) and get the serial number, then write him a "real" letter. (8:31:19 PM) sigblips: There are many other FFT libraries out there. They might not be as fast but how important is speed? (8:31:47 PM) jill_: speed is king for SETI (8:32:07 PM) avinash: But if we do conditional make, speed may not be such an issue. (8:32:10 PM) jrseti: I am assumming we will be OK here if he specifically gives us permission for our open source project. (8:32:19 PM) jill_: more speed means we can process more bandwidth in real time (8:32:20 PM) sigblips: Yes for SonATA at the ATA but how fast does the community version need to be? (8:32:42 PM) afeder: jrseti: that sounds great, also in terms of the fussiness we'd have otherwise (8:32:56 PM) jill_: i can't believe someone wouldn't think up new use that would have to be even faster (8:33:10 PM) avinash: They can use FFTW. (8:33:27 PM) avinash: It is just that we can't distribute FFTW, but people can always download it themselves. (8:33:40 PM) afeder: right (8:33:42 PM) MichaelM: good point. Make it pluggable (8:33:51 PM) jrseti: sigblips - the issue may be compatability with the FFTW3 libraries. Is there another one out there other than FFTSS? Maybe - i could not find it. (8:34:00 PM) avinash: That is what I meant by conditional Make. (8:34:02 PM) ***MichaelM belatedly notices that the license recites FFTW. duuuh (8:35:15 PM) avinash: So, 2 avenues to puruse: FFTSS, MIT. Any other suggestions? (8:35:19 PM) sigblips: jrseti: not that I know of but an interface to an FFT is fairly simple. I'm looking into plugging Apple's SSE3 FFT lib into baudline. It should be fairly easy. (8:36:40 PM) jrseti: sigblips - We can talk about this off - line. (8:36:49 PM) avinash: Offline sounds good. (8:37:09 PM) avinash: Shall we move on to the next topic? (8:37:11 PM) afeder: jrseti: how extensively is FFTW used in SonATA? could you post a quick note on what a replacement would need to do on the forum? (8:37:13 PM) afeder: avinash: okay (8:37:41 PM) avinash: Afeder - you have the "floor." (8:37:45 PM) afeder: okay (8:38:29 PM) jrseti: afeder - I have been looking at the code. To me it seems a bit too complicated to do that. But I can investigate further and see if I can explain it. (8:38:58 PM) afeder: basically, i would just like to move forward on the council/board situation .. seeing that we agreed to be an interim kind of thing (8:39:05 PM) afeder: jrseti: great (8:39:33 PM) afeder: have any of you had a chance to look at creating a charter? (8:40:12 PM) avinash: I started looking at it, but have not yet had a chance to look at other charters - you sent one from GNOME foundation. (8:40:18 PM) MichaelM: I sort of felt we were in the "what are we doing here anyway" stage (8:40:44 PM) afeder: okay (8:41:25 PM) avinash: Any other views on what we should look at before we come up with our proposal? (8:42:26 PM) afeder: i don't have any .. but a permanent kind of arrangement would be "good for business", I think (8:43:25 PM) MichaelM: +1 (8:44:18 PM) afeder: that's all :) just wanted to make sure we're still thinking about it and opposed to rolling ahead with what was originally just an interim setup (8:44:28 PM) afeder: and opposed = as opposed (8:45:51 PM) avinash: Until we finalize the charter, the CLC is interim. What we are saying is that its primary role is to come up with a permanent charter. (8:46:00 PM) afeder: right (8:46:37 PM) MichaelM: agreed (8:47:01 PM) afeder: be sure to poke us if you need any input for making the charter draft/proposal (8:47:37 PM) avinash: I will! Only one challenge - I will need some time - and I won't be on the chat next week. (8:47:46 PM) afeder: okay :) (8:48:17 PM) avinash: Some of us are out of the office Mo-We next week. (8:48:29 PM) afeder: alright (8:49:11 PM) avinash: Shall we move to the next topic? Or, is there more input on the charter? (8:49:24 PM) afeder: 1+ to moving on (8:49:33 PM) avinash: Gliese? (8:49:47 PM) afeder: sigblips? (8:50:14 PM) sigblips: Gerry Harp recently uploaded some Gliese 581 data sets. (8:50:48 PM) parutron left the room (quit: Quit: parutron). (8:51:10 PM) sigblips: Seth Shostak talks about Gliese 581g on this week's episode of Are We Alone. (8:51:59 PM) parutron entered the room. (8:52:36 PM) afeder: mhm? have questions for the data sets or? (8:53:00 PM) sigblips: I'm downloading the data now. This should be exciting. Maybe setiQuest can use all the recent news about Gliese 581 in some sort of beneficial / promotional way? (8:53:14 PM) afeder: ah (8:53:16 PM) afeder: nice point (8:53:34 PM) MichaelM: was thinking similar lines to sigblips (8:53:45 PM) afeder: except it has lost a bit of interest now with the finding being disputed and all (8:53:57 PM) avinash: Especially if the planet is there, :-). (8:54:04 PM) avinash: Afeder - beat me to the punch line. (8:54:08 PM) afeder: :P (8:54:52 PM) avinash: OK folks - anything else we can accomplish today? (8:55:09 PM) afeder: sigblips: have any specific ideas? (8:55:49 PM) afeder: we don't have much infrastructure we show off yet (8:55:55 PM) afeder: can show* (8:56:09 PM) sigblips: No specific ideas other than letting people know the data is available. Maybe some marketing genius can devise something cleverer. (8:56:29 PM) afeder: alright - let's think about it (8:56:41 PM) MichaelM: on the website, it'd be nice if the links actually went to data. most don't (8:56:52 PM) afeder: MichaelM: really? (8:57:08 PM) MichaelM: yeah, the waterfalls. hold on, i'll post the link i mean (8:57:13 PM) afeder: ah okay (8:57:19 PM) avinash: MichaelM: They do go to data, but not from the homepage. (8:57:25 PM) afeder: in most cases they don't generate waterfall plots (8:57:29 PM) avinash: And, not every data set has waterfall. (8:57:43 PM) sigblips: This one is interesting http://184.73.173.203/analysis/2010-10-01-exo-gl581_1/waterfalls.php?chan=-1257 (8:57:44 PM) afeder: the links ought to be removed though (8:57:49 PM) MichaelM: yeah, but there are links that go nowhere (8:58:10 PM) MichaelM: http://setiquest.org/join-the-quest/data-api/getting-data (8:58:20 PM) avinash: Agreed. We should at least say that no waterfall is available. (8:58:35 PM) sigblips: Not sure what it is yet. Still downloading the data. I'll let you know what I find out. Probably RFI. (8:58:43 PM) afeder: sigblips: looks cool (8:58:48 PM) MichaelM: Sandbox update? (8:59:01 PM) avinash: Yes. (8:59:44 PM) avinash: MichaelM: Were you going to give an update, or asking me for one? (8:59:53 PM) MichaelM: the files and database have been moved to the GoDaddy host. Only the front page seems to work. the rest crashes. I blame godaddy (9:00:03 PM) afeder: good strategy (9:00:04 PM) afeder: :) (9:00:50 PM) avinash: Haven't laughed this hard in a while. :-) (9:00:53 PM) MichaelM: There are a few tweaks to the godaddy host that might help, or we could get on a real server, if someone is in control of the DNS settings (9:00:59 PM) sigblips: GoDaddy? I thought you'd use 1&1 since they sponsor yacht racing. (9:01:29 PM) MichaelM: I did not pick it, and detest it! (9:01:33 PM) avinash: OK. I will go ahead and upgrade our GD service to virtual server. Hopefully that will address issues. (9:01:47 PM) avinash: Sorry folks, I am the culprit behind choosing GD. (9:02:00 PM) MichaelM: Bad Avinash! Go to your room! (9:02:03 PM) avinash: But, for my limited needs in the past they were good. (9:02:45 PM) Up2 left the room (quit: Read error: Connection reset by peer). (9:02:48 PM) MichaelM: Part of the issue is I only have limited permissions, so it's like building a ship in a bottle (9:02:50 PM) afeder: virtual server will be good, also for other things (like IRC bot) (9:02:59 PM) Up2 entered the room. (9:03:41 PM) avinash: OK. WIll try to upgrade this week, and let you know. (9:03:48 PM) afeder: neat (9:04:02 PM) sigblips: Afeder mentioned the free 12-months of Amazon cloud service on the forum. Could you use that? (9:04:50 PM) MichaelM: I'm not a new AWS client, but sure. (9:05:07 PM) afeder: technically its 750 cpu-hours times 12 or something, but it might work (9:05:28 PM) avinash: Given the donation fromAWS we don't need this free service. (9:05:40 PM) sigblips: 24 * 31 = 744 (9:05:49 PM) avinash: We have much more capacity than this free service. Can we use that instead? (9:05:58 PM) MichaelM: Dang tootin! (9:06:11 PM) MichaelM: (yes) (9:06:25 PM) afeder: sigblips: yeah, but if the server is heavily hit you might use more than 1 cpu-hour per hour (9:06:59 PM) MichaelM: I do not think the sandbox will be heavily hit at all. Or even used except in very isolated instances (9:07:09 PM) afeder: MichaelM: true (9:07:14 PM) avinash: Using AWS for the sandbox - is this another offline item? (9:07:23 PM) MichaelM: yes (9:07:26 PM) afeder: alright (9:08:10 PM) avinash: Unless there are pressing issues, should we close? (9:08:19 PM) afeder: +1 (9:08:26 PM) avinash: Remember, I am not in IRC next week - so "see you" all in 2 weeks. (9:08:35 PM) afeder: okay see you (9:08:37 PM) afeder: all (9:08:56 PM) MichaelM: likewise