#qi-hardware IRC log for Wednesday, 2011-01-26

xiangfuwe always using the name "Data Partition",  so /data is better, in fact I always using the /data then make 'ln -s /data /root'00:02
xiangfuwe can add a script file support 'mount' 'umount' 'format' data partition.00:06
xiangfuwe have one http://projects.qi-hardware.com/index.php/p/openwrt-xburst/source/tree/master/data/qi_lb60/files/usr/bin/fw_setenv_default00:06
wolfspraulxiangfu: ok /data is good, it's also compatible with jlime00:09
wolfspraulif we use /root/something or /home we just create trouble00:09
wolfspraulno symlinks either00:09
xiangfuok. I will create a package keep those nanonote special files. (those are not overwrite openwrt files)00:14
wpwrak(/home) if a user wants to, they can also just mount it over an existing /home.01:11
wolfspraulwpwrak: oh definitely. I just try to find a default way that will cause as little friction as possible, and open ways for customization.01:27
kyakxiangfu: i made a fix for gottet, not sure it'll help, but it01:27
kyakit's how it works in other apps01:28
kyakxiangfu: mplayer is playing the sample ogg file, too01:29
xiangfukyak: thanks. I will test and give some feedback .02:00
kyakxiangfu: do you know about the IGNORE_ERRORS=m flag for the make? Seems that when some pacakges fail to build (gnuplot-gfx), make skips it. But for some pacakges (guile) make doesn't skip it and exits..02:06
xiangfukyak: only the package mark "=m" ,02:07
xiangfukyak: if the package 'guile=y', it's will not continue build02:08
kyakah ok02:25
wpwraklarsc: btw, is the jz4760 already well-supported ? or would it be a major undertaking if such a chip was to be used in the next nanonote ?07:15
wpwraklarsc: (I mean on the kernel side)07:15
wpwraklarsc: particularly the asymmetric multicore worries me a bit ... but maybe the peripherals also changed a bit. seen that with samsung.07:16
lekernelFree Technology Academy Online education about Free Software: http://ftacademy.org/07:24
wpwrakphew. major infrastructure meltdown. yesterday, no elevators and only low water pressure. this morning, in addition no electricity for almost 1 hour. luckily, now things are back.08:38
wolfspraulwpwrak: about 4760 - my latest status is that yes, there are many changes, and no, nobody is planning or pushing towards cleaning up the Ingenic mess (again) and upstreaming into kernel.org etc.08:45
wpwrakwolfspraul: so quite the worst-case scenario :-(08:48
wpwrakwolfspraul: how do you like the pre-4760 XBursts ? in terms of capabilities, availability, pricing, etc. ?08:49
wpwrakwolfspraul: one problem i see is that none of them seems to support DDR. this would force you to use old memory technology, with a bad capacity/price ratio.08:52
larscyou'll have the same problem with all off them, that the peripherals are different from the jz474008:52
wpwrakoh. no reuse ?08:53
wolfspraulwpwrak: yes correct, pre 4760 has no ddr08:53
wpwraka suckish situation08:54
wolfsprauland there are many changes between them because ingenic is essentially just licensing and combining ip blocks, so stuff is quite 'jumpy' between chips08:54
wolfspraulnot suckish, that's why we have Milkymist :-)08:54
wpwrakhow do they test it ??08:54
larscsome of the more simple one are similar between SoC versions, like rtc or gpio or timers08:54
wpwraklarsc: at least something :) how about usb ?08:56
wpwrakwolfspraul: do ingenic have any ddr-cabable cpu that's not the 4760 ?08:57
zrafawolfspraul: the idea is to put the focus on Milkymist as the device for the future of qi copyleft project?.. selling it, improving it, etc?08:57
larscwell usb host is ohci anyway08:57
wpwraklarsc: ohci helps :)08:57
larscyes, the driver is 200 lines glue code08:57
wpwraka tad slow, but well ..08:57
zrafawolfspraul: since there is no an easy way to improve the current nn device without making a new one completely different?08:58
larscbut usb device has changed08:58
larscit's a OTG controller now08:59
wpwraklarsc: oh, i see. the video subsystem as well ? or is at least the "dump frame buffer" still more or less the same ?09:00
wolfspraulzrafa: don't understand your question09:01
larscit'll work i guess09:01
wolfspraulI have one product now - Ben NanoNote. And we keep trying to improve it in 20 different areas, Jlime being one of them.09:01
larscwpwrak: some registers seem to have additional fields09:01
wpwrakzrafa: i see a few issues there: 1) i don't think MM1 is mature enough yet for a proper ben-like device and will probably take a bit of time (e.g., needs an MMU, probably some optimizations, lots of power-saving work, etc.)09:01
wolfspraulI seriously start to like Jane's latest jlime-inspired crochet case, maybe I'll order one for myself :-)09:01
wolfspraulthen we are bringing out Milkymist One now09:01
wolfspraulit's still more work until it's a real product, with ce/fcc label, case, box, etc.09:02
wolfspraulthen we see09:02
wolfspraulI have no fixed plans for Ya right now, essentially it's what's being discussed on the lists.09:02
wolfspraulwe are all on the same page09:02
wolfspraulmore memory, #1 :-)09:02
wpwrakzrafa: 2) such a change would also affect higher layers. e.g., distribution efforts and such. this would mean that the whole community would restart from nearly zero. this may put off a lot of people.09:03
wolfspraulI would love better connectivity, the ben-wpan set including usb stick sounds perfect09:03
wolfspraulin parallel we see how far we can take milkymist one - that's a huge huge project09:03
wolfspraulI have no silver bullet in all this, we need to see how things are evolving, what works and what doesn't work.09:03
wpwrakzrafa: 3) such a device may be just too queer (queerer than the mm1), which would mean low sales, further increasing the financing needs09:04
wpwrakzrafa: that's why i'd hope for mm1 and a ben-like ya in parallel, and a merge maybe for the next generation09:04
wolfspraulyes I agree with wpwrak, but then it shoudl definitely be 4740-based09:05
wpwraklarsc: (fields) okay, that always happens09:05
wolfspraul64 MB SDRAM09:05
wolfspraulreuse all of the excellent kernel work we did so far09:05
wolfspraulno 50, 55, 6009:05
wolfspraulwe could even do that switchable client/host design similar to what the freerunner had, with one mini-b connector09:06
wolfspraulplus ben-wpan integrated09:06
wpwraksounds good enough to me09:06
wpwraktry to make the memory bus wider. that will already double the throughput. this ought to help with things.09:06
wolfspraulmeanwhile milkymist and m1 can mature in parallel09:06
wpwrakah, and a WQVGA :)09:07
zrafawolfspraul: about my question. I am really happy with the Ben product, and I love to help to improve it whenever I can. But I see people is trying to put the disccusion about a new product. Completely different.09:07
zrafawolfspraul: No me, again. And I am asking if they are not seeing the areas where qi project tries to improve the current Ben product.09:07
zrafawolfspraul: so my question is about a next device, on hardware side. IF the focus is on Milkymist, because it would be perhaps easier to improve on hardware side.09:08
wpwrakzrafa: i think the long-term strategy should be to have more independence on the cpu side. so the mm1 work looks very promising to me. at the same time, we have to acknowledge its limitations09:08
wolfspraulstill don't understand the question09:09
wolfspraulI can tell you my personal opinion, which is that I totally don't like jumping around building this and that.09:09
zrafawpwrak: yeah.. MM1 is another different product, no like Ben. I just was asking if maybe the idea would (because our world shows us limitations) be to focus on another product to keep live the qi project.09:09
wolfspraulthis is a copyleft project, all those people who like to talk can throw some of their savings into a little 'let's build this or that' project of their own09:09
wolfspraulof course you will see that exactly ZERO of those things will actually get built :-)09:09
wpwrakwolfspraul: (jump around) aw, i had to bite down hard on my tongue ;-)09:10
wpwrakzrafa: ah ... another product, like what ?09:10
wolfspraulit's just a freaking hard work to bring a device like Ben NanoNote together. we burnt how many people with it now? don't know09:10
wolfspraulquite a few09:10
wolfspraulnow Milkymist One, let's see whether we all survive it :-)09:10
wolfspraulso there is only a few shots we have in all this, those shots better be the right ones long term09:11
wpwrakwolfspraul: how many people did you "burn" ?09:11
wolfspraulI am very very happy with my Ben NanoNote right now, it really takes up more and more use cases slowly.09:11
zrafawolfspraul: and sorry I am not asking properly. The areas where I see Ben is improving are things I like. But I do not like much people do not see that09:11
zrafawpwrak: another product different than nn09:11
wolfspraulagreed. we need to patiently improve the device and talk about how we use it and how well it works.09:11
wolfspraulof course during the Android dark age that's not so easy :-)09:12
wolfspraulat least they are not hunting us as witches yet09:12
wpwrakzrafa: any specific ideas ? bicycle ? umbrella ? supercomputer ?09:12
wolfspraulzrafa: next product I do is Milkymist One.09:12
wolfspraul'product' as in ce/fcc, good photos, build to have in stock, press launch, etc.09:13
zrafawpwrak: no.. just the current one : real time visual effects for VJs :)09:13
wolfspraullarsc: how is Linux on your m1?09:14
wolfsprauldoes it slowly feel like a real computer to you?09:14
wpwrakwolfspraul: i think the ben per se still has a bit too many obvious shortcomings. for getting people to appreciate it more, it should be more "rounded". but yes, these could be simple improvements09:14
wpwrakzrafa: so you're suggesting to abandon the ben-ya-... line of devices and focus exclusively on mm1 ?09:15
zrafawpwrak: I would like to see more people improving the software side, no just 3, 4 guys. The same on hardware side. And I find that hard (trying to help to bring more people)09:15
zrafawpwrak: NO!09:15
wpwrakat least one clear statement ;-)09:15
wolfspraullarsc: maybe the most important Linux question is - how badly are you missing the mmu?09:16
zrafawpwrak: I like Ben.. and I am not doing any suggestions.. I just see no easy way to improve Ben (from current disscusions here) like people are liking to see (on ML mainly)09:17
wpwrakzrafa: the plan for now is: we have ben, continues to exist and to get supported. next is mm1. wolfgang can only work on making one product at a time, but that doesn't mean that the others are dead. it's just too hard to multitask such things.09:17
zrafawpwrak: yeah, I understand all of that. ANd I read here often :)09:17
wpwrakzrafa: (discussion on the list) hmm, they don't sound too unreasonable. if you ignore the usual 50% of wet dreams, then there's a good core everybody seems to agree on ;-)09:18
wpwrakzrafa: i also think the ben form factor should be preserved. it's one of the things that make it unique. otherwise, you'll quickly find yourself competing with apple and the 1001 copycats.09:19
zrafawhat is apple? :)09:21
wpwraksome foul-tasting fruit, usually inhabited by worms :)09:23
zrafawpwrak: well, and what about to continue using the current ingenic?.. will it be unavailable?.. would it hurt much because no ddr?.. would it let us to use some extra usb host?09:23
wolfspraulI think the 4740 will be available for a long time09:24
wpwrakzrafa: i think the 4720/4740 would be tolerable from a technical point of view. it is a little short on peripherals, though, e.g., it has only one MMC host.09:25
wpwrakzrafa: the lack of DDR means that we can't use modern memory. so the choices are limited, and per byte cost is higher.09:26
wpwrakzrafa: (ddr) it's not terrible yet, but it will get worse with time.09:27
wpwrakzrafa: e.g., if i look around at digi-key, SDR memory has a per byte price that's about twice that of DDR memory. (not sure if digi-key prices are really representative in this case, though. memory pricing is a bit tricky.)09:28
zrafawpwrak: ah.. I see, well I like the current Ingenic power.. so if it is there on the future and we can use it for new Bens when needed, great :) plan A09:31
zrafacurrent Ingenic = 4720/474009:31
wpwrakwolfspraul: are any of the Jz475x available ?09:35
zrafanew Bens= more ram and wpan .. ah and screen HD :)09:35
wpwrake.g. the Jz4750 seems to have dual SPI and dual MMC. that would be very welcome.09:36
wpwrakzrafa: HD3D ;-)09:37
wpwraksome sort of pointing device would be nice, too. touch screen or something else. i don't have a clear idea on that part yet.09:41
larscwolfspraul: well, it runs fine without crashing now. at least in qemu09:46
wolfspraulwpwrak: I never looked closer at 50/55 - for one there would be another big Linux effort, and then 50/55 are also far less economically successful than 40, less devices.09:46
larscwolfspraul: its not me who is missing the mmu, but a lot of programs do09:46
wpwrakwolfspraul: mmh, i see.09:47
wpwrakwolfspraul: so if going dual MMC (one inside, one outside), one would have to use GPIOs.09:48
wolfspraullarsc: do you plan to add target support for m1 to openwrt?09:52
wolfspraulor asking in a different way - how could one help with that, make it easier? I didn't follow the toolchain discussions in detail... any big work left there?09:53
larscwolfspraul: mwalle is working on adding lm32 support to uClibc09:56
wpwrak(mmc on gpio) or share the bus. but that would be unfriendly towards big-banging cards.09:56
larscand would require some work to add it proper support for it10:01
wpwrakyeah, that probably too. i also don't know how happy MMC/SD/SDIO peripherals are in general if the bus is shared.10:03
larsci guess SD/MMC would work, SDIO wouldn't at least not without some additional logic10:04
larscbut there are MMC bus share ICs10:05
larsclike http://focus.ti.com/docs/prod/folders/print/txs02612.html10:05
wpwraknow it gets really scary ;-)10:06
wpwraksuch a switch would also interfere with bitbanging cards. so better just gpio. maybe make the internal one hw-assisted, the external one gpio.10:07
larscit'll be slow. 1M/s max and the cpu will be 100%10:08
wpwraknow .. if fedex were nice enough to deliver my latest parcel, i could continue with the etching ... according to murphy's law, they door bell will ring exactly when the boards are in the acid.10:09
wpwraklarsc: yes, it'll suck. but what choices do we have ? well, maybe something like the txs02612 plus some gpios on the sdio branch.10:10
wpwrakof course, also this would need code support ...10:11
larscwhy would you need the extra gpios?10:26
larsci'll be back in 30 minutes or so...10:28
larscok, looks as if didn't miss anything...11:22
wpwrakhad a nap, to be woken up by fedex ;-)11:30
wpwraklarsc: the extra gpios for bitbanging cards11:30
wpwraklarsc: i think the ability to add simple DIY peripherals is worth preserving11:30
larscah ok11:31
larscbut i guess you wouldn't need any extra gpios for it11:33
larscwell, ok, you wouldn't be able to use the internal mmc card, if the mmc pins where used as gpio11:34
wpwraklarsc: yes. you'd want to have both options. so one approach would be to have an external multiplexer for the MMC hw engine and add regular GPIOs on the 8:10 card side of the multiplexer. if used as MMC, you'd just tri-state the GPIOs. else, you'd switch the multiplexer away from this interface and activate the GPIOs.11:36
wpwraklarsc: one limitation would be that one could do this only for one card. i.e., not both could be bitbanged. but that's probably okay.11:37
lekernel*** ERROR (IO) : Can't open a theme file specified for automatic loading19:36
kuribaslekernel: Give the full path for the theme.19:36
lekernelI used --themedir...19:37
kuribasDid you include the extension?19:37
lekernelah, works with full path19:37
lekernelno and that was the problem19:37
lekernelworks now :)19:38
lekernelah, tiling window management...19:40
kristianpaularggg i missed the interesting earlier chat :-)19:47
kristianpaulmmc have a long run of posible/coming apps, hope time will show us that19:48
wolfspraulkristianpaul: and the qi-bot and logging was down, even worse19:48
kristianpaullarsc: wich apps miss mmu, are those critical for having a sort of usefull computer?19:48
wolfsprauland wpwrak is having power problems again, so he also can't read the backlog19:48
kristianpaulwolfspraul: yeah :( i just noticed too19:48
wolfsprauloh well, sometimes everything fails...19:49
wolfspraulrjeffries: ben-wpan is not proof-of-concept22:35
wolfspraulnot sure what you mean with that term anyway. I am looking forward to using the boards in my real life.22:36
rjeffrieshi wolf. what I meant is that an L-shaped dongle off the sid eof Nanonote which (we shall see) may be knocked loose ocassinally is not what a larger potential target market would desire.22:37
kristianpauli think he pointed just the connector 8:10 not to project itself (i hope i'm right ;-))22:38
wolfspraulwejp: are you interested in gmu feedback here? I have two little ideas to improve the file browser view22:38
wolfspraulone would be that if you press enter to add a file to the playlist, the cursor advances to the next position22:38
rjeffriesIt DOES prove out the rf circuitry and assocuated low-level code, and it gets the community going with 6LoWPAN protocol22:38
wolfspraulthe other one would be to allow the selection of multiple files with shit-down, shift-up, then enter to add them all to the playlist22:39
rjeffries<kristianpaul> not ecxactly what I meant. 6LoWpan needs to be INSIDE Nanonote to appeal to a larger market22:39
wolfspraulI should send a patch but too lazy to dive into the codes now :-) or rather the way it works currently is good enough :-)22:40
kristianpaulrjeffries: define inside22:40
rjeffriesit's very clever and will sell to some percentage of Ben owners22:40
kristianpaulsoftware side is a WIP, but you cant said hw part (atben) is just proof of concept22:40
rjeffries"inside" means a new PCB that incorporates RF chip and crystal and antenna22:41
kristianpaulmy opinion22:41
kristianpaulhad you grab last commits from ben-wpan?22:41
kristianpaulbecause actually thats what mainly the pcb is22:42
rjeffries<kristianpaul> that is mot a critique, it is a statement of reality. it is brilliant work, but until it is inside the case it is a way to get Nanponote connevcted, but is not what one dreams of22:42
kristianpauli see your point22:43
wolfspraul'proof of concept' is a meaningless neologism and doesn't explain anything22:43
rjeffries<kristianpaul> we are talking past one another. I am saying TO ME it gets msuch more interesting when atBEN is integrated inside the Nanonote22:43
kristianpaulremenber when latops dint came witn wlan?22:43
kristianpaulbuilt-in wlan22:43
wolfsprauleither it works for someone or it doesn't work22:44
kristianpauli used a dongle for long time22:44
rjeffriesjesus guys, this is 2011, I just checked my calender. people expect better22:44
kristianpaulwhas that dongle prof of concept? (in that time)22:44
rjeffries<smile> I am POSITIVE you are a great person, we simple disagree22:44
wolfspraulproof of concept is just a stupid term22:45
rjeffriesthat is not relevant. we live in her and now this is not 199522:45
kristianpaulgn8 i'm off22:45
wolfspraulbecause it doesn't explain anything22:45
rjeffrieswolf it is not. it is a widely used term that many people understan22:45
wolfspraulbut of course you can continue to use it :-)22:45
wolfspraulhorrible term, really. ask 50 people and they will give you all sorts of different meanings of it.22:45
kristianpauldamn i used now, and i realize now how missunderstood can be..22:46
rjeffriesMy point is, without 6LoWpan INSIDE Nanonote, it is main;y a curiosity for a TINY band of dedicated fans. NOt to a wider audience22:46
kristianpaulused once*22:46
wolfspraulI have no idea what it means. does it work or not?22:46
rjeffrieswolf I hope you are just pulling my chain.22:46
kristianpaulrjeffries: do you own a blutooth dongle?22:46
wolfspraulno I'm very serious22:46
wolfspraulsome people confuse the main point of technology - whether it works or not22:47
wolfspraulthat's all that matters22:47
rjeffriesO I do not have a bluetooth dongle22:47
wolfspraultoday ben-wpan doesn't work22:47
rjeffriesand besides, atBen (which I am EXCITED about) will probab;ly NOT stay firmly attached to the damn 8:10 slot. There's this concept called the laws of physics. <smile>22:48
wolfspraulonce it starts to work for very strong tech people, we need to make it work easier for everybody, step by step22:48
rjeffrieswolf please don't treat me like an ideio. I know that. I think it is peachy you are doing this atben. But it is a baby step to and integated solution. dob't kid yourself.22:49
wolfspraulsure. I replied to 'proof of concept'22:49
wolfspraulyour input to this community is mostly in words, so one must be allowed to give you feedback on that at least.22:49
rjeffriess/ideio/idiot (but "ideio" does have a nice ring to it)22:50
wolfspraulyour critique is normally spot on, but you can just bring out your real point more clearly, not with such fuzzy language22:50
rjeffrieswolf I am ready to say something rude but I will bit my tounge. I sepnt quite some time a few days ago with werner and a guy from Gewrmany figuring out22:51
wolfspraulI'm not a native speaker, but that's what I picked up here and there from writers and journalists :-)22:51
wolfspraulno abstract/fuzzy language22:51
rjeffrieshow I might be able to design a PROOF OF CONCET 8:10 baord with IR on it for remote contril22:51
kristianpauloh, is it working now? :-)22:51
wolfspraulhe will tell us later22:52
kristianpaulI like IR, i think led can do some intersting stuff22:52
rjeffriesI did NOt yet taje a VOW TO YOUR cHURCH OF cOPYLEFT, BUT (sorr 4 caps)22:52
wolfspraulif and when it works, I buy one from you and try it out, follow your documentation steps on how to make it work22:53
rjeffries<kristuanpaul> is you goal to make me go awy: juest say so, and consider it done22:53
wolfspraulnot at all22:53
kristianpaulno no22:53
rjeffriesi have no dream to sell anything22:53
wolfspraulmy feedback is "proof of concept" is meaningless22:53
wolfspraulthat's all22:53
wolfspraulyou can ignore it22:53
rjeffriesjesus I simply decided what he hell maybe I'll try to build a PCB from scratch for the  pure hell of it22:53
wolfspraulthat's great!22:54
wolfspraulit will be hard though, I think22:54
rjeffriesright proving that Ben can be used with my ugly IR dongle to control the TC/Sattilite will not prove that it works. yup, totally meaningless22:55
wolfsprauldon't understand you22:55
rjeffrieswolf: do you assume my IQ is below 100? of course if I pursue it it will be a bitvh22:55
rjeffriesmy odds of sucess are low22:55
wolfspraulyou want to build an IR board for the ben to control something?22:56
rjeffriesyou guys are pissed because I never bought a Nanonte22:56
rjeffriesBFD. My friend has one, I can borrow if when I need it,22:56
rjeffriesI have not idea why I invest any time here at all.22:56
wolfspraulrjeffries: I gave you feedback on usage of 'proof of concept'22:58
wolfspraulit won't hurt but it also is not very clear to someone following the discussion22:58
wolfspraulI mean seriously - I don't know what 'proof of concept' means.22:59
wolfspraulmaybe I'm stupid22:59
wolfspraulmostly if someone says "it's just proof of concept" I would understand that as "it's not working yet, don't bother me know I'm in the middle of it, come back later"22:59
wolfspraulif a person A says "that's proof of concept" about the work of person B, I would understand it as "that stuff doesn't work"23:00
wolfspraulthere he goes23:02
wolfspraulwhen I read http://en.wikipedia.org/wiki/Proof_of_concept23:02
wolfsprauleven the entire first 3 paragraphs/lines, I cannot answer a simple question "does it work or not?"23:02
--- Thu Jan 27 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!