2011-05-07 00:21 hmm, do these looks sufficiently pr-grade ? (warning, high resolution, up to  3.5x2.6 Mpixel) http://downloads.qi-hardware.com/people/werner/ubb/vga/ubb-vga-pub-1024.jpg http://downloads.qi-hardware.com/people/werner/ubb/vga/ubb-vga-pub-v2.jpg 2011-05-07 00:22 LunaVorax_mini: btw, regarding your idea of a "reduced" ben: one problem is that removing features doesn't necessarily make a device cheaper. e.g., if you wanted to buy a ben without the keyboard, it would actually be more expensive to provide that than a regular one. 2011-05-07 00:25 pr-grade... well if you ignore the pink-white-like in letters and lines 2011-05-07 00:26 (ubb-vga-pub-v2.jpg) nice pic ! 2011-05-07 00:26 bit  scaring, but nice :-) 2011-05-07 00:28 kristianpaul: yeah, maybe i should try to reduce the red a bit ... 2011-05-07 00:37 mmh. just keeps on looking shitty. i think i'll wait for daylight and try again. 2011-05-07 00:46 or fix the color balance on the screen :) now, recharging the camera battery ... 2011-05-07 03:04 another try, less rosy this time: http://downloads.qi-hardware.com/people/werner/ubb/vga/ubb-vga-pub-1024.jpg 2011-05-07 03:06 nice :-) 2011-05-07 03:09 *hmm* now, how to upload them into the accursed wiki ? 2011-05-07 03:10 ah.. thats the easy part :) 2011-05-07 03:11 create a (dead) image link, then follow the link ? must be that, because i don't see an "upload" anywhere 2011-05-07 03:12 (upload) are you logged in? 2011-05-07 03:13 no, i don't have an account. one stupid password less to recover :) 2011-05-07 03:15 i see 2011-05-07 03:15 @#$%! "You must be logged in to upload files." :-(( 2011-05-07 03:16 actually, there's an easier way ;-) i just link to downloads.qi-hardware.com :) 2011-05-07 03:16 I will upload some of the vga pics for the news anyway 2011-05-07 03:16 the reason I haven't done the 05-01 news yet is that I keep exercising my mind over the m1 launch, or lgm press release etc. 2011-05-07 03:16 that sucks up all my story telling capacity... 2011-05-07 03:17 lgm? 2011-05-07 03:17 heh :) how are these coming along ? are the muses with you ? :) 2011-05-07 03:17 lgm = libre graphics meeting 2011-05-07 03:17 wpwrak: (link) hum, i wonder if images can be externally linked 2011-05-07 03:17 ah 2011-05-07 03:17 k 2011-05-07 03:18 Jon is planning to use/show m1, and we wanted to document that properly in a press release 2011-05-07 03:18 well I'm not complaining, I enjoy going through the process but it's hard work 2011-05-07 03:18 what we've been doing so far is not good enough, it's mostly along the lines of "this is so good that you have to love it" 2011-05-07 03:19 Milkymist tries a lot at once - new brand, new business model, new technology 2011-05-07 03:19 kristianpaul: according to http://www.mediawiki.org/wiki/Help:Images, you just put the URL. however, it says that they cannot be resized 2011-05-07 03:19 anybody who tells me "we have to ..." can just write something themselves, then it becomes obvious how hard it is 2011-05-07 03:20 the Milkymist wikipedia page was an eye-opener for me, that was good 2011-05-07 03:20 I've heard some interesting stuff stuff morning along the lines of 2011-05-07 03:20 genetics of audiovisual 2011-05-07 03:20 with this machine of hell you are in the lowest level language of visual creation 2011-05-07 03:21 like manipulating DNA 2011-05-07 03:21 hmm, "This page has been deleted". http://en.wikipedia.org/wiki/Ben_Nanonote 2011-05-07 03:22 time ago 2011-05-07 03:22 27/08/10 2011-05-07 03:23 wolfspraul: (genetics of audiovisual) sounds a bit like the things you'd find at sharism.org :) 2011-05-07 03:23 in which way? you mean it sounds like 'sharism'? 2011-05-07 03:23 ah, sharism.org - wait, checking... 2011-05-07 03:24 there are just too many ways to go about m1, it's a problem 2011-05-07 03:24 or about milkymist for that matter 2011-05-07 03:24 "it's like einstein's theory for peanut butter" :) 2011-05-07 03:24 ARM competitor? 2011-05-07 03:24 GPL licensed whatever 2011-05-07 03:24 FOSS guys, come here 2011-05-07 03:25 video manipulation? for what purpose? personal/fun, professional/office, conference, vj, art? 2011-05-07 03:25 i tought was hardware guys ;) 2011-05-07 03:25 copyleft hardware? 2011-05-07 03:25 well, as someone who doesn't presently care about the vj side (although some hands-on experience may convince me that it's a nice gadget), i see mainly the aspect of making a compact fpga-based "pc". 2011-05-07 03:25 I don't believe in being able to suddenly have a strong rally of free software folks jumping into this project 2011-05-07 03:26 yes, the dreaded fpga computer 2011-05-07 03:26 plus, it's the native habitat of llhdl, which i find a lot more interesting 2011-05-07 03:26 he 2011-05-07 03:26 like I said, this is what sucks up my story telling capacity 2011-05-07 03:26 for good or for bad 2011-05-07 03:26 and LGM is coming up in a few days... 2011-05-07 03:27 we sold NanoNotes by talking about: a) 99 USD b) Linux c) copyleft hardware 2011-05-07 03:27 but we definitely did not establish an exponentially growing anything, more like a static small community, with most members on an unsustainable path 2011-05-07 03:28 where's the breakout? 2011-05-07 03:28 and for m1, how to approach it now/ 2011-05-07 03:28 ? 2011-05-07 03:28 copyleft hw? 2011-05-07 03:28 breakout ? 2011-05-07 03:28 nah, I have doubts 2011-05-07 03:28 fpga computer? 2011-05-07 03:28 I talked to endless people about this, there is no fire/spark there 2011-05-07 03:28 about fpga computer? 2011-05-07 03:29 yes 2011-05-07 03:29 those kinds of reasonings are too dry, too academic 2011-05-07 03:29 the coolness of the fpga computer is not in the computer but in the fpga. the computer is just what enables you to make use of the fpga 2011-05-07 03:29 if Milkymist is marketed as a total fgpa-centric project, I'm sure it will fail 2011-05-07 03:29 the real use of the fpga would be for example that you can synthesize your own peripherals 2011-05-07 03:30 but if we go the 'this will be asic soon' path, that's also not good for a whole variety of reasons 2011-05-07 03:30 tough tough :-) 2011-05-07 03:30 new way computing, compiling/sinthetising the computer 2011-05-07 03:30 yes sure, but whom are we realistically targeting? 2011-05-07 03:30 hmm 2011-05-07 03:30 too early 2011-05-07 03:30 fpga makers come out with subsidized devel boards to push their latest features 2011-05-07 03:31 yes, fpga-only would be tough. also because most people don't need all that exotic multimedia stuff. 2011-05-07 03:31 sure there is zero continuity, and no products will ever emerge, but it sucks up 90% of attention for 'cool new fpga features' 2011-05-07 03:31 just as they are meant to :-) 2011-05-07 03:31 ;-) 2011-05-07 03:32 we have been largely unable to create a big excitement about free software folks, even kernel hackers, for anything below kernel 2011-05-07 03:32 so I highly doubt this angle works 2011-05-07 03:32 reconfigurable hardware video sinthesis box 2011-05-07 03:32 yes sure 2011-05-07 03:33 make me interested 2011-05-07 03:33 wolfspraul: because there's very little for them to do 2011-05-07 03:33 that sounds like a professor titling one of his classes 2011-05-07 03:33 of course he knows that his students have to attend, because otherwise they don't get enough points to pass 2011-05-07 03:33 :/ 2011-05-07 03:34 wolfspraul: strategically, i see mm1 more as a proof of concept than as a real product. alas, that doesn't fit with the planned business model. 2011-05-07 03:34 maybe I don't see the forest from all the trees... 2011-05-07 03:34 (angle) it will not work until we have better tools under kernel layer, like when llhdl gets mature.. 2011-05-07 03:34 oh no, it's a great product 2011-05-07 03:34 wolfspraul: i mean as a moneymaking product 2011-05-07 03:34 you should see where it is now, and the little details we are polishing in terms of software, electrical stability, accessories, etc. 2011-05-07 03:34 it could sell in thousands and tens of thousands 2011-05-07 03:35 wolfspraul: (k and tens of k) sounds as if you've already found your market ? :) 2011-05-07 03:35 the quality of the product is my least concern 2011-05-07 03:35 I'm on the story-telling side now 2011-05-07 03:36 i'm not talking about quality. i mean in the sense of meeting a market demand 2011-05-07 03:36 genetics of audiovisual 2011-05-07 03:36 hmm 2011-05-07 03:37 product is satisfying, I think demand can be created 2011-05-07 03:37 what is the cheapest way to create demand? 2011-05-07 03:37 who demand hardware freedom.. 2011-05-07 03:37 yes, that's one angle 2011-05-07 03:37 but I don't believe there is a strong demand there :-) 2011-05-07 03:37 instill fear in people that can only be extinguished by owning the product :) 2011-05-07 03:37 indeed 2011-05-07 03:38 neither from eff.. 2011-05-07 03:38 it's not needed either. my personal business idol here is mtk. 2011-05-07 03:38 nobody knows who they are, so what 2011-05-07 03:38 Nokia tried super hard to ignore them, pressing their eyes together harder and harder over the years. 2011-05-07 03:38 media/press has totally missed it, probably because mtk gives a damn about a marketing/pr department 2011-05-07 03:38 they focus on making money 2011-05-07 03:39 so the only place you see them is in the top silicon manufacturers charts 2011-05-07 03:39 if you don't expect "fpga" and "freedom" to sell, don't advertize them, or advertize them only in the small pring 2011-05-07 03:39 and you see how many billions of USD they make 2011-05-07 03:39 prinT 2011-05-07 03:39 if copyleft hardware functions, it will lead to superior products 2011-05-07 03:39 like many companies don't advertize that their CE devices have linux inside 2011-05-07 03:39 those products cannot be marketed as 'copyleft hardware' per se, they just happen to be 2011-05-07 03:39 but mtk have lots of turnkey products, is hard to compare i think 2011-05-07 03:40 my point is that the success of copyleft hardware does not necessarily depend on being the cool kid in town 2011-05-07 03:40 in my experience this 'cool kid in town' is all fabricated anyway 2011-05-07 03:40 it is not harder to be the cool kid in town that it is to make a ben-wpan board. it's just work. 2011-05-07 03:41 so back to m1, that's why I'm currently thinking about this 'genetics of audiovisual' 2011-05-07 03:41 I like it 2011-05-07 03:41 totally different angle 2011-05-07 03:41 wolfspraul: nice one. 2011-05-07 03:41 you can also vary the marketing message depending on audience. for a hypothetical "vj mag" you'd talk about cool effects, how it rubs this or that famous competitor in the dirt, and how rugged it is for hard stage use. 2011-05-07 03:41 also 'lowest level language of visual creation' 2011-05-07 03:41 yeah 2011-05-07 03:41 you made 'ignore marketing, make good products' sound as complicated as if sean said it ;) 2011-05-07 03:41 he, i liked that one 2011-05-07 03:41 *scnr* 2011-05-07 03:41 it's a bit wordy, but 'language' covers the programmability of it 2011-05-07 03:42 for the hypothetical "fpga forum" you'd talk about the amazingly small number of LUTs and how openrisc, sparc, etc. suck 2011-05-07 03:42 roh: well it's not that easy. we have to sell, make money. this is hardware. 2011-05-07 03:42 and so on 2011-05-07 03:42 good product, good sales force, godo marketing - needs to come together 2011-05-07 03:42 wolfspraul: if the hw and sw is great it will. 2011-05-07 03:42 if you have a bad product, marketing is too expensive 2011-05-07 03:42 I don't believe in that. there is a lot of competition. 2011-05-07 03:42 it's a factor, x*z 2011-05-07 03:42 x = product quality 2011-05-07 03:43 z = strength of sales force, including marketing message, pr 2011-05-07 03:43 exactly. have you ever seen hilti or makita advertisement banners anywhere? they dont need to. if you worked with one, you want one. no need for advertisement 2011-05-07 03:43 if your z = 0, it doesn't matter how great your x is :-) 2011-05-07 03:43 yes true, of course. I'm not arguing against that. 2011-05-07 03:43 i think they give away calendars.. so make stickers. 2011-05-07 03:44 still you have to reach someone. stories are competing. 2011-05-07 03:44 wolfspraul: sure. some salespitch needs to be. just stick to 'it does foo and it does it good' and let the rest be stuff 'just working' 2011-05-07 03:44 honest sells again ;) 2011-05-07 03:44 (lowest level language) isn't the effect programming language defined by some commercial product ? 2011-05-07 03:44 good story is not necessarily about dishonesty 2011-05-07 03:44 so you can't really claim originality there 2011-05-07 03:44 look at a movie, and movie marketing 2011-05-07 03:44 quite interesting 2011-05-07 03:44 meh. daylight. need to go home and to bed 2011-05-07 03:45 nobody needs new movies, there are plenty of old ones, great ones, that you haven't seen yet 2011-05-07 03:45 but the industry is all about creating demand for the new one, and they even invest 100, 200, even more millions of USD into it - very risky! 2011-05-07 03:45 they certainly trust their marketing power :-) 2011-05-07 03:46 wpwrak: I haven't looked much at the language yet, you mean Milkymist Patches? 2011-05-07 03:46 yes 2011-05-07 03:46 Sebastien always says it is mostly milkdrop compatible 2011-05-07 03:46 i assume that's what you refer to with 'lowest level language of visual creation', right ? 2011-05-07 03:46 that's a bsd license software 2011-05-07 03:47 (bsd) ah, nice 2011-05-07 03:47 and even if it were proprietary, programming languages are not covered by copyright, afaik 2011-05-07 03:47 yes but it's just the language similarity on m1, as a starting point 2011-05-07 03:47 I doubt there is 1 line of milkdrop code in m1 2011-05-07 03:47 me too 2011-05-07 03:48 wpwrak: that's the beauty imo. the 'lowest level language for visual creation' leaves that a bit open 2011-05-07 03:48 programming languages are tricky. e.g., they tried to keep eiffel restricted. the net effect was that it never caught on. poetic justice ;-) 2011-05-07 03:48 but is compatible, but i wonder how many people really used/use or had heard of milkydrop... 2011-05-07 03:48 it can also include the Verilog that you can equally dive into programming-wise 2011-05-07 03:49 I don't think milkdrop compatibility is an important feature in m1 2011-05-07 03:49 also wirth had this sort of ambitions with oberon, although i think it was more about keeping the language "pure". and see how this helped to spread it :) 2011-05-07 03:49 in fact if m1 grows, the language will divert from milkdrop, for sure 2011-05-07 03:49 in m1, milkdrop is just the starting point 2011-05-07 03:49 or was 2011-05-07 03:49 (it's actually a lot better than modula-2 ... almost usable :) 2011-05-07 03:49 i hope it was :) 2011-05-07 03:49 I think there are additions already today, and certainly more coming if the community grows. 2011-05-07 03:51 (compatibility) dunno. if you google for the milkdrop language, you find some activity 2011-05-07 03:51 thats good 2011-05-07 03:52 it's probably safe to say that m1 is a fork then 2011-05-07 03:52 i wonder if lekernel already wrote then about mm1.. 2011-05-07 03:52 I don't think sebastien cares that much about 100% milkdrop compatibility 2011-05-07 03:52 it's just a starting point 2011-05-07 03:52 ("language" leaving some ambiguity) yeah, i was afraid you'd aim for that :) but you still need a story why it makes sense. or be really really good at making one up on the fly if the press asks you what it means. 2011-05-07 03:53 maybe I need to play more with my m1 first 2011-05-07 03:53 what was the synthesizer name again, "interactive video synthesizer" ? 2011-05-07 03:54 I was planning to show it to a friend of mine who used to manage rock bands. he organized lots of concerts. maybe there is some spark in that conversation :-) 2011-05-07 03:54 just video synthesizer 2011-05-07 03:54 kristianpaul: would you read an article that is headlined 'genetics of audiovisual'? 2011-05-07 03:55 (show it to a musician) yeah, he'd also be able to associate with things/concepts you don't even know 2011-05-07 03:55 yes! exactly 2011-05-07 03:55 good story is good story, I have deep respect for that and the power can be big. 2011-05-07 03:55 power of good story needs to come together with power of good tech 2011-05-07 03:55 even if roh thinks that's a complicated way to say nothing :-) 2011-05-07 03:56 I'm not worried about the tech in m1 2011-05-07 03:56 almost there, almost ready for the technically unsuspecting end user. and we will finish the remaining 10% for sure. 2011-05-07 03:56 (genetics of audiovisual) .... waiting for kristianpaul's comment before i send mine .... 2011-05-07 03:59 the tech is making sure it can do what it's expected to do and keeps on doing that for a useful amount of time. i wouldn't be overly worried there either. well, the power supply situation is unfortunate. that's a weakness we already know. but i don't think it will be a major issue in a run as small as rc3. 2011-05-07 04:01 i wonder what "synthesizer" means to the average musician nowadays. e.g., would that include manipulation of live samples ? programmability ? or would they use a different term to describe such a device ? 2011-05-07 04:03 wpwrak: let's see about power supply, I always remain optimistic. It will be as good as rc2 or better. What's bad about that? 2011-05-07 04:03 I thought it's an interesting idea to translate the fpga thing as genetics, or DNA. 2011-05-07 04:04 wolfspraul: it's a bit too easy to fry the whole device, that's bad about it :) 2011-05-07 04:04 it's a translation, of course. 2011-05-07 04:04 not sure. if you plug in the power supply we ship with each unit, all is fine. 2011-05-07 04:04 yes, big IF :) 2011-05-07 04:05 well I agree with you, but let's just see whether we can do better on rc3 or not 2011-05-07 04:05 it has a nice inviting barrel connector, doesn't it ? :) 2011-05-07 04:05 the jury is still out 2011-05-07 04:05 in the big picture this is also just a small detail 2011-05-07 04:05 I'm on the story side now 2011-05-07 04:06 alright, story ... so .. to address the vj community, you need to speak their language 2011-05-07 04:06 i assume lekernel does, so that's already a good start 2011-05-07 04:07 which magazine is this? which journalist? 2011-05-07 04:08 until I see a journalist that tells me "yes this is great, I will cover this extensively in our magazine" I remain cautious 2011-05-07 04:09 I've found de-bug.de and xlr8r.com 2011-05-07 04:09 at de-bug.de I have a good intro (a journalist I know worked there years ago and still knows everybody) 2011-05-07 04:09 the 'genetics of audiovisual' just sound confused to me. genetics link closely with evolution. where's the specifically evolutionary element in what mm1 does ? e.g., does it auto-adapt to its environment ? does it procreate ? does it mutate ? 2011-05-07 04:09 of course I will try to see what they say 2011-05-07 04:10 well yes, it does most of these things, no? 2011-05-07 04:10 :-) 2011-05-07 04:10 (journalist) there, i have no idea, sorry :) 2011-05-07 04:11 how about the scene we know best? how to present it to free software people? hardware hackers? open hardware hackers? 2011-05-07 04:11 (does most of these things) well, in a sufficiently vague sense, it does everything ;-) 2011-05-07 04:12 you could probably also call it "dadaism of audiovisual", in a vague enough sense ;-) 2011-05-07 04:12 to hw hackers, i'd sell the compact core 2011-05-07 04:13 it's a fully-featured pc built around an fpga. has lots of interfaces, a good cpu, etc. so if you want to add stuff, you don't have to start from zero. 2011-05-07 04:13 of course, the mm1 hardware isn't ideal for this market 2011-05-07 04:13 devel boards pushed out for marketing reasons will take out all oxygen from that room 2011-05-07 04:14 a bit too much emphasis on multimedia, probably a bit too little on customer extensions 2011-05-07 04:14 yes, they're a very strong competitor 2011-05-07 04:15 about the only argument against them would be that they're not the reference platform this was developed on. but then, if the strength is flexibility, this argument vanishes quickly. 2011-05-07 04:16 opencores found 247 people donating money to their 'open asic project' after the coverage on slashdot 2011-05-07 04:16 so it would still be attractive for those who see themselves as pioneers. who don't want to wait half a year until someone ports it to ref board XYZ. 2011-05-07 04:16 for a total of about 8000 USD (orsoc started with 3000) 2011-05-07 04:17 or you could sell it as a long-term stable platform. but that's also tricky. probably not worth the trouble trying. 2011-05-07 04:17 that only works if the real pull is the software you are stabilizing on it 2011-05-07 04:17 Flickernoise 2011-05-07 04:17 Milkymist SoC 2011-05-07 04:17 which I believe is the correct path 2011-05-07 04:17 I think I want to stay away from competing with devel boards, as far as I can :-) 2011-05-07 04:18 (pull) also if you just want to explore. get the platform, play with the vague ideas you have, see if something sticks. 2011-05-07 04:18 I have doubts about the size of a copyleft hw pull, or the amount of free software or kernel hackers that can become deeply interesting in electronic circuits, IC design, etc. 2011-05-07 04:18 but as i said, it's not a lot of people who work like this 2011-05-07 04:19 yes those angles exist, and we should continue to be very open to them, but I don't believe there can be a real big uptake there, big growth 2011-05-07 04:19 yes, but play with what ideas? 2011-05-07 04:19 play with Verilog? 2011-05-07 04:19 play with Milkymist SoC? 2011-05-07 04:19 Sebastien is ironically doing a lot of support for people who ask him how they can get this or that part of Milkymist to work on other fpga devel boards 2011-05-07 04:19 I think i've seen it all now, Xilinx, Altera, Lattice, etc. etc. 2011-05-07 04:20 but nothing comes back, ever 2011-05-07 04:21 people may get an mm1 if they expect to work on the core itself, if the mm1 is clearly positioned as the reference platform 2011-05-07 04:21 so I think the story of Milkymist One is a story about visual effects - we have to risk to go only for what an end user can do with it 2011-05-07 04:21 maybe they will do a bit of Patches programming, maybe not 2011-05-07 04:22 maybe they will explore integration of DMX lighting systems 2011-05-07 04:22 depends a bit on their background. if they work in a reasonably well-financed project, an mm1 is quite affordable if they think it helps them 2011-05-07 04:22 then we need a separate effort to get _code contributions_ into the project, whether that's for real-time video streaming, integrating still pictures into the performance, and a myriad of other features 2011-05-07 04:22 wpwrak: I haven't seen that at all yet 2011-05-07 04:22 (only end user) yes, that seems to be the most promising direction 2011-05-07 04:22 in fact it's quite the opposite 2011-05-07 04:23 most of these people work on super small educational budgets 2011-05-07 04:23 they buy a subsidized fpga devel board for 19 USD (exaggerating) 2011-05-07 04:23 (small edu budget) ah, that's bad 2011-05-07 04:23 they have infinite time to fiddle around with their board, in fact that's the point! 2011-05-07 04:23 they will end the project in an unusable chaotic messy state, which is also OK since all those realizations were the point of their work 2011-05-07 04:24 not 1 line of code, not 1 snippet of wisdom, experience, feedback, will come back to Sebastien 2011-05-07 04:24 hmm, not sure. some of the discussions in #milkymist sound as if something was coming back 2011-05-07 04:25 be it code, ideas, experience, etc. 2011-05-07 04:26 anyway, so you don't focus marketing initially on the free or hw scene. fine. 2011-05-07 04:27 now, the end users. what do you give them ? a bag of cool effects to use for their next few gigs ? a lab to develop their own stuff ? a long-term multi-purpose platform ? 2011-05-07 04:27 very little. yes there are 2-3 guys, they have been there forever. 2011-05-07 04:27 my observation is more looking at it for 6+ months 2011-05-07 04:29 (things coming back) 2-3 guys isn't too bad. the ben is doing hardly better these days, if not worse 2011-05-07 04:29 and compare the installed base. it's enormous in comparison :) 2011-05-07 04:30 well one by one. first I'm thinking about the m1 launch now. 2011-05-07 04:30 I don't expect the contributor situation on m1 to improve fast. 2011-05-07 04:30 so maybe the only approach we should focus on is to make money for the few core contributors we have now, primarily Sebastien, and then let him hack away in happiness :-) 2011-05-07 04:31 and then maybe over time we can extend the model to more people 2011-05-07 04:31 it'll get worse in the sense that development resources will get diverted into what is basically user support 2011-05-07 04:32 so I'm mostly thinking how to sell to people who buy m1 for the visual aspect 2011-05-07 04:32 not for the freedom or programming aspect, primarily 2011-05-07 04:32 that's a nice extra for those who care 2011-05-07 04:33 yes, so  what do you give the visual customers ? a bag of cool effects to use for their next few gigs ? a lab to develop their own stuff ? a long-term multi-purpose platform ? 2011-05-07 04:33 it's simpler than that 2011-05-07 04:33 first the software will boot into rendering mode right away 2011-05-07 04:33 a sense of being ahead of the pack ? a sense of relief to have avoided falling behing ? 2011-05-07 04:33 s/behing/behind/ 2011-05-07 04:34 then we focus on two features, the simplicity is so striking that even you will be blown away :-) 2011-05-07 04:34 black and blue screen ? :) 2011-05-07 04:34 first feature is NEXT PATCH (next visual effect) 2011-05-07 04:34 second feature is 'download new free patches from Internet' 2011-05-07 04:34 :-) 2011-05-07 04:34 that's it 2011-05-07 04:35 then market that well 2011-05-07 04:35 there's a whole world to discover underneath for those who want to, all the way into the IC design :-) 2011-05-07 04:35 okay. will someone ask where those "new free patches" come from ? if you think the question will be asked, what do you answer ? 2011-05-07 04:36 e.g., does sebastien plan to spend the rest of the year writing new patches ? :) 2011-05-07 04:37 I'm sure we can organize that, the creation of new ones 2011-05-07 04:37 they are download from a url 2011-05-07 04:37 downloaded 2011-05-07 04:37 just to check, so you made your cross at the "a bag of cool effects to use for their next few gigs" option, correct ? 2011-05-07 04:37 I'm just describing the marketing approach now 2011-05-07 04:38 it's no multiple choice. no small crosses at all the other options as well :) 2011-05-07 04:38 not just gigs, for anybody who wants to dabble in this kind of visual effects/works 2011-05-07 04:38 see the box like a VGA signal generator 2011-05-07 04:38 on the box we have space to talk about 3 features, normally 2011-05-07 04:38 find. who does ? so you see a big non-vj market for shrink-wrapped effects ? 2011-05-07 04:38 hopefully the TwitterWall can fit too :-) 2011-05-07 04:38 err, for a box that plays shrink-wrapped effects 2011-05-07 04:38 1) click through xx included effects 2011-05-07 04:39 2) download new free effects from milkymist.org 2011-05-07 04:39 3) connect to any twitter hash tag 2011-05-07 04:39 implementation details :) 2011-05-07 04:39 let's talk about the usage scenario 2011-05-07 04:39 under-utilized projector around? 2011-05-07 04:39 :-) 2011-05-07 04:40 maybe we should talk to projector companies? 2011-05-07 04:40 actually a good idea! 2011-05-07 04:40 a company that sells professional projectors may also be interested in m1 2011-05-07 04:40 what would be the mix professional vj, not-quite-professional vj, hobbyist, others ? 100:10:1:0.02 ? 1:1:1:1 ? ... ? 2011-05-07 04:41 (projectors) heh, who knows. can't hurt :) maybe they'll just want to integrate it, to have a less boring "standby" display :) 2011-05-07 04:41 absolutely 2011-05-07 04:42 which would be fine by me 2011-05-07 04:42 sounds like a compfty niche, yes :) 2011-05-07 04:42 they are integrating lots of things nowadays, so you can plug in your usb stick, play videos, slide show from pictures, display microsoft office files, etc. 2011-05-07 04:43 like elphel and google. not a marriage for life, but good while it lasts. 2011-05-07 04:43 I'm also thinking about smaller companies that are more like 'systems integrators' for projectors. 2011-05-07 04:43 hmm, that sounds like office/home projectors 2011-05-07 04:43 the one in clubs may be a different caliber 2011-05-07 04:44 but an effect generator could still be interesting for them. why not. 2011-05-07 04:44 might be that they'd some major customizations, though 2011-05-07 04:44 s/they'd/they/d want/ 2011-05-07 04:45 I would think there is at least 1 layer of companies between an Acer, Sony, etc. and say a hotel anywhere in the world that needs to equip a few conference rooms 2011-05-07 04:45 no way that hotel (or hotel chain) will just order the stuff from amazon, or directly from Acer/Sony/etc. sales 2011-05-07 04:45 there must be more companies in between, and they could be interested in m1, to sell to their customers 2011-05-07 04:46 that's always a bit of a problem. company X comes and says they'd buy 10 k units if you'd just make this and this and this change. now, do you do it or not ? pro: the promise of big money. contra: they may drop out of the deal. and even if the deal gets through, it may send you doing stuff that doesn't bring you forward on your real objectives. 2011-05-07 04:47 yes but that's a normal request 2011-05-07 04:47 if it comes from a serious business, it's quickly responded to with an NRE, payment terms, etc. 2011-05-07 04:47 then they can take it or not, done 2011-05-07 04:47 could be small local companies. they buy from resellers, then take care of installation/integration. 2011-05-07 04:47 they cannot know how hard such customizations are, so the request just needs to be quickly and professionally answered 2011-05-07 04:48 okay, you're ready for that. good :) 2011-05-07 04:48 ah this is totally normal 2011-05-07 04:48 "normal" doens't necessarily imply "ready" ;-) 2011-05-07 04:49 In my former life in the professional (ahem, proprietary) industry, we've dealt with such requests from LG, Kodak, Canon, and many others. 2011-05-07 04:49 most of such requests lead nowhere, btw 2011-05-07 04:49 as in "it's normal for japan to get earthquakes and tsunamis every now and then" ;-) 2011-05-07 04:49 (nowhere) i suspected as much ... 2011-05-07 04:49 but these big companies have many managers that are exploring many ideas 2011-05-07 04:50 so they need numbers, then they run the numbers, and they find out it's too expensive/too risky/not enough profit potential 2011-05-07 04:50 so the small company protects itself against that with the NRE 2011-05-07 04:50 so is is even worth trying to compose a real response, with numbers and such ? 2011-05-07 04:50 and all is fine :-) 2011-05-07 04:51 good question, I don't know 2011-05-07 04:51 I've only seen to take this serious, yes. they are potential customers. 2011-05-07 04:51 (real response) i.e., a quote. you probably can't charge for that. 2011-05-07 04:51 which business can turn down potential customers? maybe it should close right away. 2011-05-07 04:51 qualifying customers is hard 2011-05-07 04:52 no you cannot charge for that 2011-05-07 04:53 okay, so you'll ping projector companies. do you know which ones make the big projectors for events ? 2011-05-07 04:53 (or maybe also video walls or such) 2011-05-07 04:55 and what was the expected customer mix again, pro vj/amateur vj/video enthusiast/other ? 2011-05-07 04:56 maybe we can make this easier - can pro vj and amateur vj be combined ? or are their needs too different ? 2011-05-07 04:57 also, if there's a group you feel confident just assuming to be zero, that helps with simplifying. e.g., "other" might be a candidate. 2011-05-07 04:58 (different needs) i guess that should be that needs/channels. 2011-05-07 05:01 of course, another strategy could be to focus on a known-to-be-too-small target market, just to get the thing rolling, and hope other opportunities pop up. such as those projector companies. 2011-05-07 05:02 or maybe that could be called a non-strategy ;-) 2011-05-07 05:03 professional projectors are made by Epson, Panasonic, Sanyo, Samsung, Acer, BenQ, Sony 2011-05-07 05:05 video enthousiast sounds like someone who enjoys watching movies, but we need people who like to create stuff, play with visual effects, real-time video, maybe video hacking if that exists 2011-05-07 05:05 they could have fun with m1 2011-05-07 05:07 okay, so it's not just the people who want just playback but also creators you want to reach with your marketing message ? 2011-05-07 05:08 (video enthusiast) yeah, needs a better name 2011-05-07 05:15 hmm, what's exactly the wiki dialect qi-hw speaks ? i tried to embed an image in the mediawiki style (no markup, just the url) but it doesn't take it 2011-05-07 05:18 which url? 2011-05-07 05:18 it's a standard mediawiki 2011-05-07 05:18 no dialect :-) 2011-05-07 05:18 http://downloads.qi-hardware.com/people/werner/ubb/vga/ubb-vga-pub-1024-511x383.jpg and http://downloads.qi-hardware.com/people/werner/ubb/vga/ubb-vga-pub-v2-377x383.jpg 2011-05-07 05:18 m1 is not about playback, it's about realtime manipulation of a video _signal_ 2011-05-07 05:19 (mm1) yeah .. effect playback :) 2011-05-07 05:19 features like twitter feed are new, or maybe the embedding and manipulation of still images 2011-05-07 05:19 or the streaming of the generated output 2011-05-07 05:19 but that's a lot of work and should not be part of the message today 2011-05-07 05:19 and not needed imo 2011-05-07 05:19 what we have is good and flexible enough 2011-05-07 05:19 also, the manipulation aspect is still a bit downplayed, compared to the pure synthesis aspect 2011-05-07 05:28 Milkymist as a standby signal for a projector 2011-05-07 05:28 I like it 2011-05-07 05:29 those somewhat respectless ideas may surprise... :-) 2011-05-07 05:29 if they can put nearly the equivalent of a ben into a glossy magazine, ... ;-) 2011-05-07 05:33 I just like the idea of emailing those corps saying "I want to sell a new standby signal for your projectors to you" 2011-05-07 05:33 at least someone will laugh and look at that :-) 2011-05-07 05:34 should also be good for the press :) 2011-05-07 05:42 how many presentable patches are there at the moment ? enough to mention their number in the initial marketing message ? 2011-05-07 05:45 not sure, I need to pay more attention to the patches 2011-05-07 05:45 yes, there are a lot and they are nice 2011-05-07 05:46 along the lines of the standby signal, one could also say m1 is a nice way to fill time between talks, at a conference 2011-05-07 05:46 ;-))) 2011-05-07 05:46 so instead of making m1 the central thing, sell it as a way to fill time 'in between' 2011-05-07 05:46 why not. if it sells... 2011-05-07 05:46 and there is indeed _a lot_ of time between talks, at every event 2011-05-07 05:47 i just though that it could also nicely replace a lot of boring presentations at company meetings. dilbert may like it :) 2011-05-07 05:47 with video-in patches the audience could be engaged, with a hash tag they could display things etc. 2011-05-07 05:47 maybe not even start the next talk because it's too much fun :-) 2011-05-07 05:47 when the next guy wants to start, he's booed off stage. we should make a promotional clip along those lines... 2011-05-07 05:47 who hasn't been at a boring talk??? 2011-05-07 05:48 this could go so wrong. imagine response to audio input. before you know, you'll turn a reasonably quiet if bored audience into a roaring mob ;-)) 2011-05-07 05:49 now we have something for rejon to set up. place camera. upload to youtube :) 2011-05-07 05:52 hmm. how many simultaneous audio inputs does it have ? 2011-05-07 05:59 simultaneous? 2011-05-07 05:59 there is line-in, plus microphone 2011-05-07 05:59 the rest would have to come over Ethernet (not implemented now), or USB (even more effort), or hardware hacks I guess 2011-05-07 06:00 so two right now, line-in + mic 2011-05-07 06:00 mic is the built-in one ? 2011-05-07 06:01 line-in is two channels, left and right ? 2011-05-07 06:01 good question, yes I think it's stereo 2011-05-07 06:01 mic is built-in, yes 2011-05-07 06:02 line-in is typically connected to a music player 2011-05-07 06:03 okay, with stereo you could do this: place one external microphone to left, one to the right. place the camera in the middle, looking at the audience, ideally from above but not too high. analyze the audio signal for peaks and determine the difference in arrival time between left and right 2011-05-07 06:04 and then? 2011-05-07 06:04 :-) 2011-05-07 06:04 sounds like a fun start 2011-05-07 06:04 when you get a peak, digital-zoom in at the origin of the peak. run some wacky effect on it for a few seconds. then zoom out again. 2011-05-07 06:05 ok that kind of programmable camera interaction would be new though 2011-05-07 06:05 needs a motor controlled camera, zoom, etc. 2011-05-07 06:05 but of course, it all exists 2011-05-07 06:05 no no .. all static 2011-05-07 06:05 take an image of the room and digitally zoom 2011-05-07 06:06 only the pixels move, not the equipment 2011-05-07 06:06 well, until things get seriously out of control ;-) 2011-05-07 06:07 hmm, we may just have invented the successor of karaoke. needs even less skills ;-)) 2011-05-07 06:10 only drawback is that this would already block all the audio inputs. well, you could have an external switch/mixer 2011-05-07 06:11 what's the maximum video in resolution ? vga ? 2011-05-07 06:11 another good question, don't know 2011-05-07 06:12 about zooming in automatically to 'moving' parts, that sounds like a serious undertaking as well 2011-05-07 06:12 of course this is doable, but remember we have more or less one full-power engineer working on the whole stack right now 2011-05-07 06:14 the video signal is pal or ntsc, doesn't that answer the resolution question? 2011-05-07 06:14 you basically need two components: one that analyzes the audio and returns a peak yes/no and peak position indication in real time, plus the zoom. 2011-05-07 06:14 (one engineer) yeah, that's a bit limiting ... 2011-05-07 06:15 hmm, sounds vga-ish or worse then 2011-05-07 06:15 and it's not clear yet whether we can support 3 different video signals or not (with multiplexing) 2011-05-07 06:15 so for now let's assume only one 2011-05-07 06:15 one video would be sufficient 2011-05-07 06:15 it's just the crowdcam. the rest is generated inside 2011-05-07 06:16 well, you could also have multiple cameras if you want to get fancy 2011-05-07 07:14 impressive .. imagemagick can convert xcf files, too 2011-05-07 07:33 wolfspraul: http://forums.winamp.com/forumdisplay.php?f=84 except that we suck at doing the same 2011-05-07 07:52 [commit] Werner Almesberger: ubb-vga/web/: generate a Web page describing UBB-VGA (im progress) http://qi-hw.com/p/ben-blinkenlights/64e74ab 2011-05-07 07:52 and the result looks like this: http://downloads.qi-hardware.com/people/werner/ubb/vga/web/ 2011-05-07 07:55 with (almost completely) automatic image generation from the schematics, and so on. feel the power of my Makefile, mediaweenie ! ;-) 2011-05-07 08:17 [commit] Werner Almesberger: ubb-vga/web/index.html: some changes in wording and text structure http://qi-hw.com/p/ben-blinkenlights/3f054e1 2011-05-07 08:23 [commit] Werner Almesberger: ubb-vga/web/index.html: minor wording improvement; give images more space http://qi-hw.com/p/ben-blinkenlights/0937c4a 2011-05-07 09:08 lekernel: don't understand we suck at what? 2011-05-07 09:08 running a forum? running those particular patches on m1? 2011-05-07 09:57 morning 2011-05-07 09:58 just submitted a slashdot story about the UBB-VGA hack. 2011-05-07 09:58 not sure whether it gets posted, but for those who have an account, you might be able to up-vote it 2011-05-07 09:58 http://slashdot.org/recent 2011-05-07 09:58 "Micro-SD Card Slot Abused as VGA-Port" 2011-05-07 09:59 http://slashdot.org/submission/1566326/Micro-SD-Card-Slot-Abused-as-VGA-Port 2011-05-07 10:00 (just an experiment, first article i ever submitted) 2011-05-07 10:03 wolfspraul: making the project popular and having people submit patches 2011-05-07 10:04 dvdk: cool 2011-05-07 10:06 guess the slashdot moderator's are still sleeping, maybe shouldn't have submitted that early "in the morning" :) 2011-05-07 12:43 oh shit. now i need to fix my image links **quickly** :) 2011-05-07 12:45 dvdk: very nice article, by the way. concise yet says everything that needs saying. 2011-05-07 12:46 wpwrak: thx 2011-05-07 12:47 btw got a mail from the slashdot system some 30m ago saying it was posted.  but as it hasn't shown up yet, so either "posted" isn't what i think it means, or there's some delay built in 2011-05-07 12:49 keeps hitting reload :) 2011-05-07 12:51 "not so fast" said the girl to the sailor ... 2011-05-07 12:52 ah, ok, it looks like it's really coming.  http://hardware.slashdot.org/story/11/05/07/1215246/Micro-SD-Card-Slot-Abused-As-VGA-Port 2011-05-07 12:52 tagged as 'hardhack' :) 2011-05-07 12:53 still not linked from the front-page but they have some delay in there so that paying subscribers get stories earlier. 2011-05-07 12:54 wolfspraul: (genetics) i personally dont read articles with such us tittle 2011-05-07 12:56 [commit] Werner Almesberger: ubb-vga/web/: link to medium-size images, not full size; some rewording http://qi-hw.com/p/ben-blinkenlights/91bc5ea 2011-05-07 12:56 sorry wpwrak i got asleep yday :/ 2011-05-07 12:56 phew. now they can come 2011-05-07 12:57 kristianpaul: (genetics) that's exactly what i though about it too ! :) 2011-05-07 12:57 wpwrak: wrt your last commit: trying to brace for the slashdot effect? 2011-05-07 12:57 i dont get the sense/context of the genetics word in that title 2011-05-07 12:58 dvdk: naw, saving them from a nasty 2.5 MB surprise when they click on an image ;-) 2011-05-07 13:01 s/m1 is a nice way to fill time between talks/m1 is a nice way to fill walls 2011-05-07 13:03 people just dont know what they can do, you need tell it how, in a short video or something 2011-05-07 13:03 even me,, i discove trought xiangfu that _awesome_ patches to keyboard feature 2011-05-07 13:03 so a video about how to do that is not bad 2011-05-07 13:04 dvdk: hmm, there should also be some link for further communication. well, let's see if people can figure it out on their own. 2011-05-07 13:05 my point is an article is not enought, i will head more to whatever writing poiting a demostration of the current milkymist features 2011-05-07 13:06 wich btw i dont get yet how the patch editor can be really usefull if you dont have a window to edit it separatelly, recompile and get fun with it 2011-05-07 13:07 kristianpaul: the number of articles you can get out this way is limited. so you need a second level with more information, and connect it to these articles. 2011-05-07 13:07 oh sure, but thats when things get hard ;) 2011-05-07 13:08 btw about mic, will be nice to know wich patches reacts with the mic-in i have the "feeling" some do and others dont, but you get catught when it works, if youy we're making a sound and watching the wall with the milkydrop effect 2011-05-07 13:10 kristianpaul: put a LED next to each input and turn on the LEDs if the input is being used ? 2011-05-07 13:10 wpwrak: nice 2011-05-07 13:10 kristianpaul: video, mic, line in, ... 2011-05-07 13:14 dvdk: I don't think every post, even if accepted in a category, makes it to the homepage 2011-05-07 13:18 kewl. works. 2011-05-07 13:19 shall i just link from the bottom of the web page to the qi-hw list and this channel ? 2011-05-07 13:20 wpwrak: yes i think the qi-hw list is pretty important.  was not so easy for me to find, when i "accidentally" bought my nanonote. 2011-05-07 13:22 done 2011-05-07 13:22 [commit] Werner Almesberger: ubb-vga/web/index.html: added links to #qi-hardware and the mailing list http://qi-hw.com/p/ben-blinkenlights/f1efa9e 2011-05-07 13:28 if it does get onto the homepage, I would expect about 8000 visitors on the first day, then maybe 5000 on the second day, then drop a lot 2011-05-07 13:28 at least those were roughly the numbers last time the NanoNote was on slashdot 2011-05-07 13:28 back then that converted into about 100 NanoNote sales in the US and Canada alone 2011-05-07 13:29 seems that we made the front page. brace for impact :-) 2011-05-07 13:29 unfortunately the server couldn't really hold up well, maybe visitors (and sales) would have been a bit higher with a more scalable system 2011-05-07 13:29 let's hope the all-static content on downloads holds up better ;-) 2011-05-07 13:30 slashdot is just moderated now? they don't have their voting system anymore? 2011-05-07 13:30 still voting I think, though I wouldn't hold my breadth for the democracy aspect there 2011-05-07 13:30 Chinese democracy? :-) 2011-05-07 13:30 ah hold on, that will be called 'democracy with Chinese characteristics' when they implement it 2011-05-07 13:30 :-) 2011-05-07 13:32 I see the nanonote is on the /. frontpage 2011-05-07 13:33 feels a little silly for having reused rejon's introduction line that put my name before anything else 2011-05-07 13:34 kodein: great 2011-05-07 13:35 now this should happen once a month or so :) 2011-05-07 13:35 yeah 2011-05-07 13:35 man I'm too slow 2011-05-07 13:35 I will put an infobox right on the top talking about Milkymist too :-) (on the wiki page it links to) 2011-05-07 13:36 lekernel: we have wpan, which should be rolling off the factory belt any moment now. if they haven't already. (i hope not ... for i don't think anyone is prepared to actually test the critters) 2011-05-07 13:37 he nice 2011-05-07 13:37 oh, hace a hardhack tag.. 2011-05-07 13:40 ah hard is for hardware.. 2011-05-07 13:40 wpwrak: good 2011-05-07 13:40 so, for next month we already have one story in the pipeline 2011-05-07 13:40 we need a slashdot article for milkymist :-) 2011-05-07 13:41 sure, we are working on some press releases in parallel 2011-05-07 13:41 I can't access slashdot.org ... works for you? 2011-05-07 13:41 slashdot articles for milkymist are easy (ok, and not easy). Easy in the sense that there are at least 10 different slashdot stories in it. 2011-05-07 13:42 and not easy in the sense that they need a gifted writer, and the right idea/headline, to make a good post for slashdot 2011-05-07 13:42 it needs to be digested slowly, so we actually need a series of milkymist stories :-) 2011-05-07 13:42 well the lua/dmx hack the other dat may applied?.. 2011-05-07 13:42 or what should be postend on /. 2011-05-07 13:42 ? 2011-05-07 13:43 stories, i see 2011-05-07 13:43 well yes, but it needs to be put in context 2011-05-07 13:43 it needs a good headline 2011-05-07 13:43 it needs the right timing, i.e. it has to be done multiple times over time, and so on 2011-05-07 13:43 David did great, he just did it :-) 2011-05-07 13:43 wolfspraul: /. may not be such a bad place for spreading the word about MM1. there, you can probably squeeze in multiple articles about different angles of the same product more easily than at traditional places. 2011-05-07 13:43 it's just one form away - same is true for pretty much every other online publication btw 2011-05-07 13:44 'tip us' 2011-05-07 13:44 they want others to write the good texts 2011-05-07 13:44 ok I posted a bit desperate sounding banner on the top of the page 2011-05-07 13:44 http://en.qi-hardware.com/wiki/Ben_NanoNote 2011-05-07 13:44 feel free to edit if you think it's too tasteless :-) 2011-05-07 13:45 I don't expect a storm of viewers anyway, the headline we have on slashdot is a bit scary... 2011-05-07 13:45 he. qi-hardware is taking a bit to load.. 2011-05-07 13:45 here too 2011-05-07 13:45 "Connection to Server Refused" 2011-05-07 13:45 yeah well 2011-05-07 13:45 it loaded, but no pics.. 2011-05-07 13:45 good thing I just got the banner uploaded 2011-05-07 13:45 slahdot effect ! 2011-05-07 13:46 wow how desesperate was that  ;) 2011-05-07 13:46 he :-) 2011-05-07 13:46 http://downloads.qi-hardware.com/people/werner/ubb/vga/web/ still loads. all hail static content ! ;-) 2011-05-07 13:46 editable by anyone. I just did a quick hack, with a smile hopefully people get that... 2011-05-07 13:46 hehe 2011-05-07 13:46 cool, yes, we made it. 2011-05-07 13:46 dvdk: :-) 2011-05-07 13:47 still, slow load.. 2011-05-07 13:47 maybe it'd be a good idea if some of us helped clear up misconsceptions coming up in the comments 2011-05-07 13:47 dvdk: congratulations ! and thanks !! 2011-05-07 13:47 but yes surelly better than a wiki 2011-05-07 13:47 oops 2011-05-07 13:47 "La conexión tardó demasiado tiempo" 2011-05-07 13:47 nice ! 2011-05-07 13:47 wpwrapk: nothing to thank for, it's still all your hard work 2011-05-07 13:47 wolfspraul: argh. sound a bit too desperate indeed. 2011-05-07 13:47 edit it 2011-05-07 13:47 or give me a better text 2011-05-07 13:48 I change urgently to 'maybe' :-) 2011-05-07 13:48 "hey check out last coming product" 2011-05-07 13:48 first step to friendliness... 2011-05-07 13:48 much better :) 2011-05-07 13:48 replace "buy" wth "yoy can get" ? 2011-05-07 13:49 (good if you do the editing. here it's too glacially slow :) 2011-05-07 13:49 first the server has to let me through 2011-05-07 13:49 wolfspraul: is fast !! 2011-05-07 13:49 banner is gone.. 2011-05-07 13:49 really? maybe someone didn't like it? :-) 2011-05-07 13:49 hehe 2011-05-07 13:49 surelly 2011-05-07 13:50 actually I hope there are no spam attacks just now, that would be kind of mean 2011-05-07 13:50 hum, that probably to be expected 2011-05-07 13:50 spammers should watch high-traffic blogs and magazines, and then attack the urls of sites that are just appearing on those sites 2011-05-07 13:51 wpwrak: I changed to friendlier wording. 2011-05-07 13:51 just dont put slahdot on the header 2011-05-07 13:51 i think 2011-05-07 13:51 oh maybe someone thinks this is spam actually :-) 2011-05-07 13:52 lol 2011-05-07 13:52 I'm trying to see the page history, will know in a sec 2011-05-07 13:52 wolfspraul: i think you need a prioritized vpn ;-) 2011-05-07 13:52 worst case I turn the admin flag on for the page 2011-05-07 13:52 better just do that :) 2011-05-07 13:53 i foudn this checking my rss/history for qi-hw http://en.qi-hardware.com/wiki/File%3AProduction_PCB_atusb06.jpg 2011-05-07 13:53 no there was no edit. kristianpaul can you reload? 2011-05-07 13:53 the new banner should be there 2011-05-07 13:53 ah.. 2011-05-07 13:53 is not loading here.. 2011-05-07 13:53 should I set the admin flag on? 2011-05-07 13:53 maybe I just do it for peace of mind 2011-05-07 13:53 second 2011-05-07 13:57 still loading.. 2011-05-07 13:57 may be is just me and been far away from germany ;) 2011-05-07 13:58 i guess the cgi based wiki is currently hitting its limits 2011-05-07 13:59 yeah, now i see why people at wikimedia uses proxies :-) 2011-05-07 13:59 funny it finally loaded and no banner 2011-05-07 13:59 anyway.. 2011-05-07 14:00 you cannot compare the typical mediawiki install with the custom setup used by wikipedia.org etc 2011-05-07 14:00 sure :) 2011-05-07 14:00 kristianpaul: what do you mean with 'no banner'? 2011-05-07 14:00 in fact they are doing a big service to the world by keeping the same software used on that site, and still make it easily installable and usable by everybody else 2011-05-07 14:00 it's a lot of extra work 2011-05-07 14:00 dvdk: wolfspraul put a banner/announcement in a red like box in the top qi-hw 2011-05-07 14:01 the bottlenecks on a typical machine like the one that hosts qi-hardware.com starts with memory (the way Apache is setup), and then probably already CPU 2011-05-07 14:01 or may be is gone now already.. 2011-05-07 14:01 kristianpaul: if that's "the banner" than i can see it when loaded from here (germany) 2011-05-07 14:01 dvdk: yeah I am shamelessly trying to steal some peoples attention and tell them about Milkymist 2011-05-07 14:01 hope you don't mind... 2011-05-07 14:01 kristianpaul: "Welcome Slashdot readers -..." 2011-05-07 14:01 wolfspraul: it's all your problem now :) 2011-05-07 14:01 dvdk: not here 2011-05-07 14:02 kristianpaul: then your browser/proxy might do some caching. 2011-05-07 14:02 wolfspraul: yes it reads a little desperate 2011-05-07 14:02 still! 2011-05-07 14:02 that's already after wpwrak toned it down 2011-05-07 14:02 probably 2011-05-07 14:03 dvdk: can you suggest a more noble wording? 2011-05-07 14:04 as i said dont use the word Slashdot on it 2011-05-07 14:04 and surelly will sound less desesperate ;) 2011-05-07 14:05 ok. that page is super chaotic anyway. so a bit more chaos at the top in a banner won't hurt, indeed. 2011-05-07 14:05 I remove that welcome thing 2011-05-07 14:05 wolfspraul: "we have been slashdotted.  to keep you from further stressing this wiki, let us just point you to the Milkymist (our next-gen project) and Tuxbrain+Sharism (our distributors for EU resp. the rest of the world) 2011-05-07 14:06 dvdk: nice ! 2011-05-07 14:06 just an idea.  have to go now.  as i said, it's your problem now :) 2011-05-07 14:06 he he 2011-05-07 14:06 lol 2011-05-07 14:06 yes, very good text 2011-05-07 14:06 will do 2011-05-07 14:09 wolfspraul: (chaotic) don't you say ! ;-) i think that wiki needs three things: a) a more hacker-friendly foundation, e.g., git-based (i've been whining about this for a long time, i know :) b) better performance. not sure where the specific bottleneck is, though. c) serious cleanup. some pages are hopelessly overloaded while others don't quite show enough and/or are dead ends. i think a) could also help to better control those dead en 2011-05-07 14:09 ds. 2011-05-07 14:10 wolfspraul: for better consistency, we'd need to find (a) wiki editor(s). not sure if anyone is volunteering. there are people who love to do this kind of work, but i'm not sure if many of those are hanging out around here. 2011-05-07 14:14 can't edit anymore, oh well 2011-05-07 14:15 keep on trying :) 2011-05-07 14:15 did you set the admin flag ? 2011-05-07 14:17 yes, already done earlier 2011-05-07 14:17 sd... sd... 2011-05-07 14:17 :) 2011-05-07 14:18 kristianpaul: ? 2011-05-07 14:18 wpwrak: remenbering the SD-like branding 2011-05-07 14:19 microSD ans such 2011-05-07 14:19 ah yes. i carefully avoided that one ;-) 2011-05-07 14:20 used MMC a few times, though, but i hope that's reasonably safe 2011-05-07 14:30 wow, you're on slashdot :D 2011-05-07 14:32 17:00  17:30 2011-05-07 14:32 Fabricating a Libre Graphics Future 2011-05-07 14:32 Jon Phillips 2011-05-07 14:32 Day 4  Friday May 13 2011-05-07 14:32 I hope thats mm1 related talk.. 2011-05-07 14:32 wow just before closing talk 2011-05-07 14:35 no there is no mm1 talk 2011-05-07 14:35 ok 2011-05-07 14:35 Jon was planning to use the m1 in breaks and such 2011-05-07 14:35 nice 2011-05-07 14:35 well, if it really happens 2011-05-07 14:35 that's the problem with the press release. if I am not sure what is really happening there, I will not push out any press release. 2011-05-07 14:36 he, thats always a problem if you dont carry with you right equipoment like projector... 2011-05-07 14:36 so I will write a draft, and then hopefully we get a few more facts to fill in 2011-05-07 14:36 if it's credible, push it out to go through the motions 2011-05-07 14:37 if not, then we rather wait until we have some real news, not just make up a fake story that didn't even happen in reality. That would be quite embarassing :-) 2011-05-07 14:44 wolfspraul: well, if rejon assures you that he'll do something specific (whatever it is he plans to end up doing), you can probably announce it at face value. then maybe make a nice after the event report once it has occurred. 2011-05-07 14:46 comments on /. are good so far. some who don't get it, others who explain them, one troll, ... good catch so far :) 2011-05-07 15:09 http://xkcd.com/409/ :p 2011-05-07 15:10 ;-)) 2011-05-07 15:16 the "to keep you from further stressing this wiki" also has a bit of an odd ring to it. for the next wave, maybe we should have something along the lines of "exploring this site may be slow for a while. you may find these interesting:" 2011-05-07 15:16 or similar. still doesn't sound too great. 2011-05-07 15:23 I still can't access slashdot 2011-05-07 15:27 qi hardware server stats:  11:27:20 up 100 days, 14:53,  3 users,  load average: 50.83, 46.43, 41.89 2011-05-07 15:27 well... 2011-05-07 15:27 :D 2011-05-07 15:28 i've never seen a load avg that high 2011-05-07 15:29 wait no, you'll wake up roh !! 2011-05-07 15:29 ;) 2011-05-07 15:55 it's all over twitter: http://twitter.com/#!/search/nanonote 2011-05-07 15:56 ok good, we need more of this stuff 2011-05-07 16:16 wolfsprau1l "genetics of audiovisual" is a meaningless phrase. 2011-05-07 16:17 anybody else noticed, that the irc-log hasn't been updated during the last few hours? 2011-05-07 16:25 hmm 2011-05-07 16:25 qi-bot: alive?? 2011-05-07 16:25 dvdk: seems that all is fine.. Note the the log is int UTC 2011-05-07 16:25 qi-bot: ping 2011-05-07 16:25 ah yes UTC is consuding at first 2011-05-07 16:26 seem fine 2011-05-07 16:27 good wpwrak is goint to FISL too, 2011-05-07 16:27 last intro mail sent.. 2011-05-07 16:28 waht is involved with creating a new patch for milkymist? 2011-05-07 16:28 kyak: now i get the latest log.  hmm, maybe a caching issue overe here. 2011-05-07 16:28 rjeffries: a bit of RTFM about milkydrop, 2011-05-07 16:29 sadly :/ 2011-05-07 16:29 i dint fully understand it yet.. 2011-05-07 16:34 kristianpaul wolfsprau1l making it relatively wasy for a milkymist user to create and modify patches using high level controls sounds like a genuine need 2011-05-07 16:35 dvdk the irc log seemed to be unreachable at first for me. then is mysteriously came to life 2011-05-07 16:36 rjeffries: not so, people can do minor modification and learn/experiment from there 2011-05-07 16:36 there are lots of patches ready to work 2011-05-07 16:37 you dont need start from scratch allways 2011-05-07 16:37 unless you really know what you want to do :) 2011-05-07 16:38 dvdk: i think the irc log is updated only about every 30 minutes. with the shashdot hordes hammering at the server, it may be a bit slower :) 2011-05-07 16:40 (twitter) nice :) 2011-05-07 16:40 rjeffries: as you can see on http://forums.winamp.com/forumdisplay.php?forumid=84 people actually manage to write these things 2011-05-07 16:40 doing it in Flickernoise isn't any harder 2011-05-07 16:41 plus we're going to have an online repository of pre-written patches with single-click update, for the lazy and clueless 2011-05-07 16:42 hum, seems wpwrak is also replying some comments at ?. 2011-05-07 16:42 lekernel: i hope your fanbase loves it when you talk dirty to them ;-) 2011-05-07 16:43 kristianpaul: just one so far 2011-05-07 16:43 yeah i knew it was you :) 2011-05-07 16:45 kyak: I run an open, unencrypted WiFi access-point, firewall WiFi off from my wired network, and run OpenVPN on top of the WiFi link if I need satellite-to-home encryption or unfiltered access to my wired network. 2011-05-07 16:46 kyak: `open but still encrypted to the AP' isn't really all that great an idea, 2011-05-07 16:46 wpwrak I just pimped your VGA thingie on Identica and Google Buzz 2011-05-07 16:49 rjeffries: great, thanks ! 2011-05-07 16:49 let's see how long until NYT picks it up :) 2011-05-07 16:49 kyak: ... because, while you [might] be insulated from the other users of the AP, you still end up trusting the AP and its operators. 2011-05-07 16:50 It's end-to-end or bust. 2011-05-07 16:51 rozzin: you are reading too much logs :) 2011-05-07 16:51 I think there actually *was* a movement to do `open but encrypted to the AP' a decade ago.... 2011-05-07 16:52 wolfsprau1l what hardwrae is sed for your server. it is acrting hammered. a good problem to have, 2011-05-07 16:52 Or maybe I'm just remembering the move by hotspot-providers to `secure' things. 2011-05-07 16:54 I vaguely remember seeing a `why having a `secure' link to an AP operated by an unknown/untrusted operator *doesn't* provide *you* with security' article, about ten years ago. 2011-05-07 16:55 kyak: I'm actually just slow to respond. 2011-05-07 16:55 kyak: I'll tell you more about *that* in a week or so... ;) 2011-05-07 16:56 rozzin: hehe :) 2011-05-07 16:57 rozzin: you are building a VPN tunnel to your wired network, do i understand correctly? What should "simple" users do when they accessed your opened, unencrypted WiFI? 2011-05-07 16:58 kyak: You understand that correctly. 2011-05-07 16:59 kyak: `simple' users don't get access to my private resources on my wired network. period. 2011-05-07 17:00 kyak: That's why those resources are on a wired (physically-based access) network in the first place ;) 2011-05-07 17:02 kyak: People who are using my AP to go upstream should do the same thing regardless of whether the short/near link was encrypted: *don't trust me* not to snoop the packets passing through my AP. 2011-05-07 17:03 [commit] Werner Almesberger: index.html: moved author's name; file format of images we can load was incorrect http://qi-hw.com/p/ben-blinkenlights/88e09d0 2011-05-07 17:04 kyak: They shouldn't trust my ISP, either. 2011-05-07 17:06 kyak: It's amazing what you can see when you control the router--I've done `I can see what you're looking at online even before you do' demonstrations for houseguests before. 2011-05-07 17:08 rozzin: have you hard of driftnet ? :) http://www.ex-parrot.com/~chris/driftnet/ 2011-05-07 17:08 My wife had a friend over, and one of them must have said something that lead me to explain network-topologies.... 2011-05-07 17:09 Oh, I think one of them must have asked why it was so slow, and then watched me boot a freeloader. 2011-05-07 17:09 ... who was running bittorrent or something. 2011-05-07 17:10 s/hard/heard/  # argh 2011-05-07 17:10 Then the obvious question was asked: `how did you do that, or even find that out?' 2011-05-07 17:11 wpwrak: ... so I started up driftnet... 2011-05-07 17:11 ... waited a few seconds, and then asked, "So, which of you is looking at the picture of the samurai'". 2011-05-07 17:14 Neither of them knew what I was talking about, until 10 minutes later--when that image finally, actually appeared on the friend's screen (she was on some javascript-heavy site that was doing aggressive preloading). 2011-05-07 17:15 they would be so proud at minority report ;-) 2011-05-07 17:15 rozzin: if only everyone had their own private VPn server :) 2011-05-07 17:16 She yelled, "Wait--there's the samurai! WTF!? How did you do that!?" 2011-05-07 17:17 So then I got out the network-diagram again.... 2011-05-07 17:17 And they both went, `ooooh...'. 2011-05-07 17:18 kyak: Well, the point is that I'm *not* going through the VPN most of the time. 2011-05-07 17:19 kyak: I've used it something like 3 times in the past 6 years. 2011-05-07 17:19 kyak: because most of the Internet is outside my VPN anyway--see? ;) 2011-05-07 17:19 well.. guess i just have more dirty secrets 2011-05-07 17:20 kyak: Even if I passed through the VPN, I'd still have to go over an unsecured link run by an unknown/untrusted operator to get to the Internet. 2011-05-07 17:21 kyak: Several of them, in fact--try running `ping -R' or traceroute, some time :) 2011-05-07 17:22 kyak: Like I said, it's *end-to-end* or bust. 2011-05-07 17:22 rozzin: that's for sure, but it's not the reason to neglect your security. 2011-05-07 17:22 at least where you can 2011-05-07 17:22 kyak: Hunh? 2011-05-07 17:24 i doubt some provider overthere is analyzing the whole traffic passing by. It's more probable that someone in your neighbourhood is doing that 2011-05-07 17:24 so you should take care about encryption in those parts where you have control 2011-05-07 17:24 that's the best you can do, and it's better than nothing 2011-05-07 17:25 kyak: If you reduce the number of points at which someone can evesdrop on your online banking session from 10 points to 9, most of the sense of secuurity there is false. 2011-05-07 17:25 kyak: If you just secure the link end-to-end, you are better off-- 2011-05-07 17:27 i just hope you are not suggesting to drop the security at all because of that many points your traffic is passing through 2011-05-07 17:27 and adding additional layers of encrypted comms with local AP-operators (who you just said are more likely to be the ones sniffing your packets!;)) doesn't buy you anything beyond what the end-to-end HTTPS link gets you. 2011-05-07 17:28 kyak: I think you need to lookup what "end-to-end" means :) 2011-05-07 17:29 http://www.publicityinsider.com/release.asp 2011-05-07 17:29 actually analyzing the whole traffic is a service you have to pay if want security :p 2011-05-07 17:29 ha 2011-05-07 17:29 kyak: what I'm saying is: if you care about security, use https instead of http or ftp, use ssh instead of telnet, etc. 2011-05-07 17:30 kyak: *do not trust local AP-operators like me*. 2011-05-07 17:30 he 2011-05-07 17:32 rozzin: i'll write it down as an advice of the year, thanks :) 2011-05-07 17:33 dont use the internet !! 2011-05-07 17:34 kyak: ... because I've already revealed that I'm snooping on your traffic--regardless of whether you encrypt the link *between* us ;) 2011-05-07 17:36 I definitely would not trust the 19-year-old running the AP at the coffee-shop. 2011-05-07 17:39 There's no reason to expect them to have had any ethics education at all, so I'd basically expect thinks like the infamous Zuckerberg IM session :) 2011-05-07 17:41 I'd probaly trust a 30-year-old running the cafe's AP less than the 19-year-old, too :) 2011-05-07 18:30 Is there some sort of audio-recorder that works well on Ben NanoNote? 2011-05-07 18:32 arecord 2011-05-07 18:32 arecord -d 8 -f cd -t wav test.wav 2011-05-07 18:33 aplay later.. 2011-05-07 18:33 Mm. I was thinking something less... command-line :) 2011-05-07 18:33 oh, well just install jlime 2011-05-07 18:34 http://en.qi-hardware.com/wiki/Jlime 2011-05-07 18:34 I don't want jlime, just a recorder-app! 2011-05-07 18:34 well... 2011-05-07 18:35 shrugs. 2011-05-07 18:36 jlime includes a recoder app btw 2011-05-07 18:36 i wanted to mean that 2011-05-07 18:36 Yeah. 2011-05-07 18:36 you are free to remove the other appps.. 2011-05-07 18:36 I'll have to look at what that is. 2011-05-07 18:36 please :) 2011-05-07 18:37 gotta go, bye 2011-05-07 18:38 I'm building an OpenWRT-based system for my wife, and she would like to be able to use the microphone to record the noises that the baby makes. 2011-05-07 18:38 But it'd be an auxiliary function. 2011-05-07 18:40 The primary use is to run ncmpc on vt1, and mpdjay in the background. 2011-05-07 18:42 GJay needs all of the CPU & RAM it can get, so I guess I'm looking for something almost as slim as arecord :) 2011-05-07 18:43 Hm. Build failed... 2011-05-07 18:43 Oh, right--foxtrot. 2011-05-07 19:22 mildly interesting 2011-05-07 19:22 http://www.measurementdevices.com/index.php?name=News&file=article&sid=2405 2011-05-07 20:03 rfid? 2011-05-07 20:04 sounds surprising they managed to make a decent 13MHz antenna in such a small space 2011-05-07 20:58 hmm, is qi-hw down ? 2011-05-07 21:03 well, turandot at least (projects/downloads) 2011-05-07 21:07 ah, just very very very slow .... 2011-05-07 21:26 :D 2011-05-07 21:28 heh, turandot:  17:11:33 up 100 days, 20:37,  2 users,  load average: 89.75, 86.85, 89.29 2011-05-07 21:28 no surprise it's slow 2011-05-07 21:28 what's abusing it? :) 2011-05-07 21:28 hmm 2011-05-07 21:28 wpwrak I supose :D 2011-05-07 21:30 question: is the website 'Qi' down? no reaction, there 2011-05-07 21:30 use to read the irclogs there ;) 2011-05-07 21:30 it's been slashdotted. seems that the evening horse is worse than the morning horde ;-) 2011-05-07 21:30 in the morning, we got "only" a load average of around 50 2011-05-07 21:31 hmm, 107 mysqls daemons running on turandot. and 59+11 apaches 2011-05-07 21:32 the main problem seems to be memory, though 2011-05-07 21:32 ah .. 9 apache2-git combos. maybe they're the greedy ones 2011-05-07 21:34 let's see if i can get a "ps" ... the response time feels like those good old batch-oriented computers. 2011-05-07 21:34 sources for an 32mb machine so big?? 2011-05-07 21:35 checks Slashdot. 2011-05-07 21:38 Mmm. 2011-05-07 21:38 linux IS a batchOS :P 2011-05-07 21:38 well, at least it came from there ;) 2011-05-07 21:43 wb botty 2011-05-07 21:43 memory seems to be mainly used by a number of  git ls-tree 2011-05-07 21:43 wpwrak: hmm.. the bot is running under "eggdrop" user 2011-05-07 21:44 kyak: eggdrop seems to behave 2011-05-07 21:46 load average: 80.07, 89.18, 90.40   getting better 2011-05-07 21:47 users can't get served and give up 2011-05-07 21:48 yeah, probably. 2011-05-07 21:49 roh: any quick load control measures you can recommend ? seems that apache shouldn't accept all that many connections. 2011-05-07 21:49 roh: (i suppose you have root on turandot ?) 2011-05-07 21:50 load average: 69.25, 80.53, 86.72 2011-05-07 21:50 feeling almost responsive :) 2011-05-07 21:52 there should be MaxClients in apache config 2011-05-07 21:52 might be something else 2011-05-07 21:52 ListenBacklog 2011-05-07 21:59 not sure if listenbacklog makes a bit difference. that would be more like kernel memory wasted on connections you'll never serve. once that is the problem, you're pretty much dead already anyway ;-) 2011-05-07 21:59 i'll stop apache for a oment 2011-05-07 21:59 MaxKeepAliveRequests !=0 (o is infinite) 2011-05-07 21:59 such a racist thing to do ... at the first sign of trouble, kill the natives 2011-05-07 21:59 http://paulgazis.com/Humor/Vikings.htm 2011-05-07 21:59 MaxClients defaults to 256 2011-05-07 21:59 apache2 stop is running ... it takes a long time dying, writhing in agony 2011-05-07 21:59 that's about 200 more than the machine can handle :) 2011-05-07 22:03 mass extinction event still in progress ... 2011-05-07 22:03 uptime crept up to 80.99 again 2011-05-07 22:03 err, load average 2011-05-07 22:03 apache is still a patchy server ;) 2011-05-07 22:04 btw : does anyone knows of an shell-app for android?? *g* 2011-05-07 22:06 Fusin: if you are talking about remote shell, that would be connectbot 2011-05-07 22:07 i think this shutdown will take a good while. taking a shower ... 2011-05-07 22:07 nah, i think of an app working like an xterm, invoking bash or /bin/sh :) 2011-05-07 22:11 has an usb keyboard connected to the android tablet 2011-05-07 22:11 there is some default shell app in cyanogenmod, i'm not sure how it's called.. it's not convenient anyway, i use adb 2011-05-07 22:11 is now more like a subnotebook 2011-05-07 22:22 Die Hard 2011: The Apache  (still not dead) 2011-05-07 22:25 he's gone ! finally :) 2011-05-07 22:28 load average: 18.99, 50.12, 71.59 2011-05-07 22:29 still a few processes busy catching up ... 2011-05-07 22:29 MaxClients lowered from 150 to 40 2011-05-07 22:29 General Custur did it faster earlier ;) 2011-05-07 22:30 *Custer (aka Hard-Backsides) :D 2011-05-07 22:30 planetplanet seems to be raking the disk. kjournald and pdflush both quite busy. python running under "list" as well 2011-05-07 22:32 Custer dies trying ... historians quibble a bit about what exactly ;-) 2011-05-07 22:33 only one planetplanet vs. mysqld now. and, and a bit of python ana php5 2011-05-07 22:33 i'll give it a bit more time to settle down. getting my laundry now ... 2011-05-07 22:35 wpwrak: turandot? 2011-05-07 22:50 roh: yes 2011-05-07 22:51 i lowered the MaxClients from 150 to 40. should i do anything else before i restart apache ? 2011-05-07 22:51 machine is now idle. all the other (backlogged) tasks have completed as well. 2011-05-07 22:54 well, let's just see what happens :) 2011-05-07 22:54 an army of apaches starts ... 2011-05-07 22:55 some of them have git as a child 2011-05-07 22:55 wpwrak: dunno that machine.. 2011-05-07 22:56 [commit] Werner Almesberger: ubb-vga/web/index.html: correction - it's the response we fake, not the "busy" http://qi-hw.com/p/ben-blinkenlights/19c0440 2011-05-07 22:56 good. also git commits work again :) 2011-05-07 22:59 wolfspraul: i took down turandot's apache server for about half an hour and lowered MaxClients from an overly optimistic 150 to 40. load average had gone through the roof, >80 2011-05-07 23:00 wolfspraul: restarted it a few minutes ago. so far, it's behaving 2011-05-07 23:01 ahh, I was just wondering 2011-05-07 23:02 yes 40 is good 2011-05-07 23:02 is it a virtual server? 2011-05-07 23:02 the Piwik stats show ca. 5100 visitors so far 2011-05-07 23:03 how many was it during the last invasion ? 2011-05-07 23:03 yes, virtual 2011-05-07 23:03 wpwrak: whats running on that machine? seving http only should be easy for some hundred connections at a time 2011-05-07 23:04 roh: lot of things. planet, lists too, it think. downloads, some mystery mysql. git. 2011-05-07 23:04 there's tons of stuff on it http://en.qi-hardware.com/wiki/Server_setup 2011-05-07 23:04 virtual servers are always tricky, no ones really knows what the 100 other virtual webmasters runs... 2011-05-07 23:05 planet is a bit of a hog. kept on running for some 5-10 minutes after apache was dead. then mysql took another 5+ more to calm down. 2011-05-07 23:05 I do because the entire physical box is rented by me. 2011-05-07 23:06 wpwrak: the most imporant thing is that the connections are closed fast, successfully 2011-05-07 23:07 lesson for the next PR campaign: put the web server on its own machine and make sure all content linked in the announcement is static ;-) 2011-05-07 23:07 planet runs from cron in regular intervals 2011-05-07 23:07 not that evil. 2011-05-07 23:07 renders static output 2011-05-07 23:07 roh: now apache seems to run smoothly. i think the problem with planet was that several jobs were piling up 2011-05-07 23:08 that should never happen (multiple runs interleaving) 2011-05-07 23:09 yes, correct. connections need to be served fast or rejected right away. 2011-05-07 23:09 i saw several (3-4, don't remember exactly) planet processes burning ~60-70 CPU among each other. 2011-05-07 23:09 so reducing maxclient from 150 to 40 was correct, although 40 maybe is a little low now. 2011-05-07 23:09 the entire machine can be optimized in many ways, but that's all work in itself and I don't want to end up with 10 different machines, say openmoko style :-) 2011-05-07 23:10 yeah, i picked a conservative limit. see if there are any surprises :) i think the git-to-web interface is also a bit "expensive" 2011-05-07 23:10 to prepare for better scalability, I rather increase memory, or move to a bigger box, or even just make this machine the only one on the physical host 2011-05-07 23:11 wolfspraul: openmoko were not many machines. just more vms. makes administration much easier (no intersections where not neccessary) 2011-05-07 23:12 I know but I doubt that and I don't want many vms. 2011-05-07 23:12 it may make sense to separate potential high-load services from the more predictable ones. that way, when a wave hits, you only have one set of services to worry about and not fires popping up all over the place. 2011-05-07 23:13 anyway, it's all your choice ;-) i marked the two changes in apache2.conf with " - wa" 2011-05-07 23:14 http://en.qi-hardware.com/wiki/Server_setup#Apache2 2011-05-07 23:14 it was 70 before, if the documentation is still correct 2011-05-07 23:15 i found 150 at both places 2011-05-07 23:16 there were about 70 apaches running at the time the machine nearly choked. so 70 wouldn't be such a bad guess :) 2011-05-07 23:16 apache2-+-61*[apache2] 2011-05-07 23:17      `-9*[apache2---git] 2011-05-07 23:17 and   `-mysqld---105*[{mysqld}] 2011-05-07 23:17 it was 70 (see that config file), and 70 was too much 2011-05-07 23:17 he, run 'mysqltuner' 2011-05-07 23:18 the Apache config file is in /etc/apache2/conf.d/qi 2011-05-07 23:18 ah, i edited /etc/apache2/apache2.conf 2011-05-07 23:19 lol 2011-05-07 23:20 I will adjust some of those mysql values now... 2011-05-07 23:25 wpwrak: the problem may also have been mysql 2011-05-07 23:26 maybe something got stuck there which in turn got the Apache clients stuck 2011-05-07 23:27 yes, could be. mysql kept on running for a very long time after i started stopping apache 2011-05-07 23:28 (the stopping of apache itself took some 10-15 minutes) 2011-05-07 23:48 ok, /me gosub sleep 2011-05-07 23:48 cu 2011-05-07 23:49 http://www.guardian.co.uk/technology/2011/apr/30/apple-chinese-workers-treated-inhumanely 2011-05-07 23:51 yes all sounds correct 2011-05-07 23:53 I like how these NGOs (or this article) stays relatively down to earth and just reports facts. So it helps people understand each other, which is good. 2011-05-07 23:54 I mean the 2 NGOs specifically mentioned as the source of these 'findings'. 2011-05-07 23:55 kristianpaul: in that list, which item you find is the most surprising, or most 'inhumane'? 2011-05-07 23:58 wolfspraul: btw.. one feat of virtualisation is: seperated ressource limiting ,) 2011-05-07 23:58 may be the excessive overtime but that happene everywhere. 2011-05-07 23:58 let the download apache not take down the wiki etc. 2011-05-07 23:59 the "confession letter" looks odd to me