#qi-hardware IRC log for Thursday, 2013-01-03

sagexhey guys I am new00:07
Action: sagex expects to be hazed00:07
whitequarkwe prefer to avoid that and get to the work instead.00:13
jianjuGreets, all. Having trouble booting from USB on a Dell 17R SE 7720 running Windows 8 with a secure boot partition. Can anyone shed some light on the subject?00:18
whitequarkwpwrak: http://developer.mips.com/clang-llvm/00:54
whitequarkyou might be interested in trying it out 00:54
wpwrakoh, thanks, but i'm very comfortable with gcc so far :)00:59
wpwrakyou know, with the years, that certain anxiousness to reject everything that's been around disappears a bit as some things have earned your trust, and you've learned to not pay attention to a lot more, too01:00
wpwrakjianju: i find your problem description to be amazingly self-explaining. windows 8 plus "secure' boot with usb added for fun... and you're having booting problems :)01:02
wpwrakjianju: alas, you're unlikely to find help for your problem in this channel as most people here stay well clear of just those causes01:03
wpwraksagex: now that you mention it, we actually have some depolarized bits piling up in /dev/null that could use some cleaning ...01:04
whitequarkwpwrak: you clearly didn't see much of gcc internals ;)01:05
whitequarkon the other hand, you don't exactly need to hack on them, either.01:05
wpwraksagex: but there's actually no single dominant topic here, so the fact of you being here, new or old, doesn't tell us all that much01:05
wpwrakwhitequark: the fact that i've been rather successful in avoiding to do that in the last 20 years or so suggests that you may be right :)01:06
whitequarkwpwrak: ... or that you were avoiding that precisely because they were so arcane01:07
wpwrakwell, 20+ years. damn, how time flies01:07
whitequarks/were/are/01:07
wpwrakoh, gcc has a certain reputation :)01:07
kristianpaullets hope it keep it01:08
wpwrakbut then, i know enough about compilers that i don't presume i'll really understand what one is doing without spending a considerable amount of time on it01:08
wpwrakso if the compiler actually happens to add obscurity on top of intrinsic complexity, that actually helps :)01:08
whitequarkwpwrak: (anxiousness to reject everything) sometimes I think that it would be cool to replace every thing GNU for the same reasons as GCC01:09
whitequarkbut then I very clearly understand that it would bring even more compatibility problems01:09
wpwrakah, and "one" refers to a specific implementation. i've written "real" compilers before, so i know the drill01:09
kristianpaulreal compilers for what language btw?01:10
whitequarkwpwrak: (compilers) certain codebases have this intrinsic property that you can look at it and more or less understand what a certain self-contained module does, and how01:10
jianjuwprak: thx 4 replying all the same. I took a chance that someone here would know the trick. I'll keep looking.01:11
whitequarkwpwrak: LLVM is one of them. this is the main reason I like it, and not the age of either of us ;)01:11
wpwraknaw, don't fight the gnu. most of the gnu-ish things are pretty stable and not overly offending. we even got used to --option, even though everbody hated it back then01:11
Action: kristianpaul loves the not very maintained gnu screen01:12
wpwrakkristianpaul: one i created myself, called "Cora". some weird mixture of my impressions of pascal, C, and forth-ish languages01:12
kristianpaulforth ah :)01:12
whitequarkwpwrak: (don't fight the gnu) exactly. it took me some time to become accustomed to it. also, --option is awesome--I can't work in BSD's at all :)01:12
whitequarkkristianpaul: some people prefer newer alternative, tmux01:12
wpwrakkristianpaul: ran on cp/m where it beat turbo pascal in code density :) i also ported it to ms-dos but there, the competition was stiffer01:12
kristianpaulwhitequark: yeah sure but as it works as good as my procmail :-)01:13
Action: whitequark uses postfix...01:13
kristianpaulwpwrak: ah you really like/learnt to fight chip resources constrainsts dont you? :-)01:16
whitequarkwpwrak: it seems to me that at least in certain parts of IT, newcomers mainly learn the most recent new cool thing01:16
wpwrakkristianpaul: my later forays into compiler/language design didn't involve traditional code generation, so they probably don't count. i'm rather proud of my umlsim language, though, which adds object orientation in an almost clean way to C :)01:16
whitequarkwpwrak: ten years ago in webdev it was PHP. five years ago Ruby. now it is Node.JS, which is oh god horrible01:17
wpwrakwhitequark: and they fight the old stuff, for obvious reasons :)01:17
whitequarkwpwrak: not only fight, but also copy 80% of it at the very beginning01:18
wpwrakwhitequark: much like oldtimers dislike the new stuff, also for obvious reasons :)01:18
whitequarkwpwrak: it is sometimes hard to understand if you dislike something new because it's new or because it's actually horrible01:18
wpwrakwhitequark: yeah, copy 80% until it works, then add 20% fresh problems :)01:18
whitequarkI can't stand node.js. it was created by professionally unfit people. on the other hand, rust (which is even newer), is really good IMO.01:19
whitequarkfor example.01:19
wpwraki think in many cases things are basically equivalent. so there's no gain in actually learning the new thing. plus you have a lot of existing work that won't cooperate with it.01:19
wpwraki don't even know what "node.js" is supposed to be. i've heard of javascript, though :)01:20
whitequark(learning) a whole lot of new things is actually a reiteration of old things. now functional programming is cool again. if your intent is to become a better programmer, you can as well return to '80s and read some papers.01:20
whitequarknode.js is basically a standalone javascript engine which is supposed to work as a webserver01:20
whitequarknote that javascript is strictly single-threaded, so you are confined to prefork. also you get all other "good" parts of JS, most importantly the fact that everything is done with callbacks.01:21
whitequarkso all the processes which are commonly written in a blocking way are suddenly non-blocking, because js requires you to do it01:21
wpwrakkristianpaul: well, those chip constraints were in the bad old BASIC days. it still makes me kinda feel guilty to need more than a handful of kB for a program. back then, your imagination was the limit of your universe, but its diameter was always 1 kB (ZX81), later 3.5 kB (VIC-20).01:22
whitequarkwpwrak: and you're stuck with mile-long callback chains. and don't forget errbacks!01:22
wpwrakinteresting. a language optimized for writing bugs ?01:22
whitequarkbesides that, it doesn't actually introduce any new or even slightly interesting concepts, except for the fact that you now *might* run identical code in your browser and on your server01:23
whitequarklike, for example, validations for forms01:23
whitequarkbecause you, of course, can validate everything without a roundtrip to the server. like username uniqueness.01:23
whitequarkwpwrak: there's also MongoDB, a NoSQL database promoted by roughly the same kind of people01:24
wpwraki love those client-side validations. so much fun to read those security issue bulletins ...01:24
whitequarkby default, it returns _before_ it actually writes the data to disk, _to achieve better results at benchmarks_ in order to bash relational database01:24
whitequark*databases01:24
wpwraklet's just say that, as the universe gets older, the amount of noise increases :)01:25
whitequarkthis is just so wrong on so many levels I don't even know where to start :)01:25
whitequarkit is also quite fun to watch people praising NoSQL as the new big thing, given that BerkleyDB actually predates relational databases01:25
whitequark(there are very nice and useful nosql solutions, but for 99% you need an rdbms/only rdbms)01:26
wpwraknaw, even unit write() returns before data is written to disk. normally, that's okay. what's a bit troubling is that it's hard to insert barriers.01:26
whitequarkwpwrak: relational databases issue such barriers before returning from a transaction. of course.01:26
wpwrakheh, yes, or postgres01:26
whitequarkthis is the whole point of transaction. and you need to go additional lengths to ensure your journaling FS won't munge that01:26
whitequarkpostgres is RDBMS :)01:26
wpwraksure. and it's *ancient*01:27
wpwrakbesides, the oldest database is still the file system. a lot of people seem to forget that.01:28
whitequarkwell, not exactly ancient, but surely not new01:28
whitequarkwe use postgres everywhere in our webdev company and <3 it01:29
wpwrakof course, it lacks the benefit of vendor lock-in, since you can just poke around with ls, cat, rm, and friends01:29
whitequarkhehe01:29
wpwrak(postgres) kewl. it's nice that it's getting some love even today.01:29
whitequarkit's basically what is used everywhere except the simplest sites (for which mysql is ok too)01:30
whitequarkand it still improves quite a bit01:30
whitequarkwpwrak: (SNR ratio) my solution to keeping with that is stay sane is as follows: look at every single new cool thing. learn it. evaluate it by technical merit, but never, ever, by userbase. try to keep my opinion with myself. don't switch unless it has actual tangible benefit.01:32
whitequarks/is/and/01:33
wpwraknice, but eventually you'll get bored with the new thing of the week01:33
whitequarkwpwrak: not sure I understand you here01:34
wpwrak"hey look, they're doing that mistake from 5.18 years ago again" :)01:34
whitequarkoh, let me explain it another way01:34
wpwrakso with time you look for additional competence indicators01:34
whitequark"know something? don't like it? don't give a fuck, it will go away by itself."01:34
whitequarkmost certainly.01:34
whitequarknot knowing it might be actually dangerous, because good new things tend to eat your own pie01:35
whitequarkbut if they're not good... who cares01:35
wpwrak"good" alone is not enough. unless you're looking for a cause to fight for, it still needs to come with a somewhat broad acceptance.01:36
wpwrakand again, with time, you're less eager to find causes to fight for.01:36
kristianpaulwpwrak: nice limits, well thats actually even less what implement a cpu on a fpga with built ram so still room for creativity dont you think? 01:36
wpwraknot because a lack of energy (well, sometimes too) but simply because eventually you solve or reach a truce with the most pressing need or issues01:37
wpwrakso the number of fights to pick shrinks01:37
wpwrakkristianpaul: those fpgas have excessive resources ;-)01:38
whitequarkwpwrak: I don't agree with the acceptance part01:38
whitequarkwhen Rails took off, it was a fringe framework in a very, very fringe language01:38
kristianpaulwpwrak: exactly :-D01:38
whitequarkfrom japan01:38
whitequarkafter Rails took off, there is now a Rails clone in every single language, and that's certainly a good thing01:39
wpwrakkristianpaul: alas, it also shows in the price. i would hope to see wolfgang's nice work also to have an effect on cplds. they're often a lot more accessible, yet they're rarely seen in the open hardware scene.01:39
kristianpaulwpwrak: i think was azonenberg who said was going to work on cheap cplds01:40
wpwrakit's indeed impressive that an idea from japan makes it. that doesn't happen very often, for a number of reasons. but lack of originality usually isn't one of them.01:41
wpwrakkristianpaul: kewl01:41
wpwrakkristianpaul: soon we'll have to solve the fpga board problem01:41
kristianpaulwpwrak: the bga thing?01:41
whitequarkwpwrak: (japan) there are communication problems even right now, but they're not threatening to the ecosystem or anything like that01:42
wpwrakkristianpaul: well, maybe ...01:42
wpwrakkristianpaul: seems that wolfgang isn't very concerned with dealing with the potential of his work just yet01:42
whitequarkwpwrak: I heard that with other japanese projects, there was some degree of nationalism, and it didn't help at all. Basically original developers expected everyone else to treat them as higher-ups and not peers. Very fortunately, and somewhat unexpectedly, none of the Ruby folks do that. This is probably the main reason Ruby is where it is now.01:43
kristianpaulwpwrak: well one single man work dont get time to realize all that potential :)01:43
wpwrakkristianpaul: in particular, it would be a great seed for a community. but that community will need hardware to play with.01:43
kristianpaulwpwrak: hardware and facilities01:43
wpwrakkristianpaul: xiangfu's bga boards could work for that, but you'd have to produce them since bga is diy-hostile01:44
kristianpaulwpwrak: not all people will want to understand chip foorplanning for just making a blinking led...01:44
kristianpaulwpwrak: indeedn xiangfu's board are just right, anyway you could use a single bus for the rest of the comunication01:44
wpwrakkristianpaul: i'm not talking about all people. i'm talking about those who understand what this means. obviously, we're not millions :)01:44
kristianpaulwpwrak: ahmm =)01:45
wpwrakkristianpaul: nevertheless, this is significant work. he's forcing open a door where there was previously a wall. with the right spin, this could be great.01:46
wpwrakconsider linux - back then, the whole hacker world was ready. and then linus provided the seed. he's great, but he couldn't have done everything himself. but he didn't have to, since lots of people had been waiting already for this opportunity.01:47
wpwrakand of course, not "people on the street" but capable developers01:48
whitequarkwpwrak: it's way easier with software01:48
kristianpauland people with business ideas :-)01:48
wpwrakxiangfu's boars have the problem that you need to industrially produce them. they're non-diy. not sure if wolfgang is prepared to do this.01:49
wpwrakwhitequark: yes, software is always easier :)01:49
whitequarkwpwrak: besides, things with fpgatools (I assume you're talking about them) look way more interesting in this discussion than before, for me01:49
whitequarkwhere can I read more about it?01:49
wpwrakwhitequark: software, been there, did it, we won. no on to the next challenge, hardware :)01:50
wpwraks/no/now/01:50
kristianpaulwpwrak: but he could use other friendly slx9 models but to be honest bga will soon or later there for diy01:50
kristianpaullater be there*01:50
wpwrakkristianpaul: i don't think so. bga is inherently unfriendly because you can't inspect it.01:51
whitequarkwpwrak: (we won) yeah! and I think that it is the right time to go after it.01:51
kristianpaulwpwrak: yes sure01:51
whitequarkwpwrak: what about manufacturing BGA modules which expose all pads? this isn't that expensive yet is suitable for diy perfectly01:51
wpwrakwhitequark: you mean, besides the source ? :) i don't think there is anything else01:51
whitequarkwpwrak: discussion maybe? ok, source it is :)01:51
wpwrakwe don't need "BGA adapters". the chip in question is available in a QFP package01:52
whitequarkah01:52
kristianpaulas soon as those slx9 QFP still avaliable i dont see problem either01:53
wpwrakit's just that no fully DIY-able design compatible with our needs exists yet01:53
kristianpaulanyway bistreams should be same no?01:53
kristianpaulxiangfu: Hi01:53
wpwrakalso, there is an LX9 board that's not too expensive: http://www.em.avnet.com/en-us/design/drc/Pages/Xilinx-Spartan-6-FPGA-LX9-MicroBoard.aspx01:53
wpwrakvaries between USD 90-120, depending on where you are01:54
whitequarkwpwrak: ohhh, C only. I would even agree on using Python, but only C? not friendly, especially for prototyping.01:54
wpwrakC is perfectly suitable for this01:54
wpwrakbesides, this is a rock-bottom low-level interface. you don't want to start wasting your cycles already there01:54
wpwraki mean, what's next ? memcpy written in Common LISP ?01:55
kristianpaulhum?01:55
whitequarkwpwrak: 1) you would be correct if the interface and its goals were set in stone. but the project is in very early stage.01:55
kristianpaulwpwrak: yes nice board still remember is not real price and is subsided some how01:55
whitequark2) memcpy in common lisp doesn't inherently have to be slow. there is plenty of evidence to prove that, including, I suspect, literal memcpy implementations in CL.01:56
wpwrakkristianpaul: that's not the issue :)01:56
wpwrakwhitequark: if you memcpy in CL is fast, it's not native :)01:56
whitequarkwpwrak: why?01:57
wpwrakkristianpaul: the issue is that, once fpgatools go public, people will want to join the effort. maybe just to play. maybe to contribute.01:57
kristianpaulwpwrak: yes correct01:57
wpwrakkristianpaul: now, what hardware do they use ? build a frigging BGA board ? no way. design their own ? maybe. buy something ? yes, most would like that.01:57
whitequarkwpwrak: as per examples: PyPy for Python, Maxine VM for Java. all of the above are meta-circular VMs written completely in their target language and yet achieving native speed on operations like memcpy.01:58
wpwrakkristianpaul: so there's a point in the continuum and that's the avnet board. available, has the LX9, not too expensive, but has certain issues.01:58
whitequarkwpwrak: besides, that's not what I'm trying to communicate at all.01:58
xiangfukristianpaul: Hi01:59
kristianpaulxiangfu: hey :)01:59
kristianpaulmorning there01:59
whitequarkwpwrak: if you can hack on something while wasting 10x less time on the quirks of your language, it's worth whatever bad runtime performance if it's not literally hours. history proved that countless times.01:59
xiangfuwpwrak: I improved: http://en.qi-hardware.com/wiki/Mini-slx9#mini-slx9-qfp144. 01:59
wpwrakwhitequark: yeah, the thing is that the runtime of certain languages isn't suitable for such low-level operations. so you out them out of the box. kinda like gcc builtins.02:00
whitequarkwpwrak: _when_ your project matures and _if_ you actually need that speed, you simply go and rewrite it in C. it's really easy once you have an architecture which is known to work02:00
xiangfuwpwrak: the plan was switch bga with fab-made pcb. since I think I have enough experience on home made pcb.02:00
whitequarkwpwrak: and you can trivially replace it part by part with all modern dynamic languages.02:00
kristianpaulxiangfu: oh nice wfp board02:00
wpwrakwhitequark: you'd be surprised by the number of people who are extremely comfortable with C ;-)02:00
xiangfukristianpaul: good morning02:00
hellekinwpwrak: you forgot your cups :)02:01
whitequarkwpwrak: again, not what I'm trying to communicate here02:01
wpwrakxiangfu: mmh, now that's a lovely sight :-)02:01
whitequarkwpwrak: I'm comfortable with C. I just don't see the point to do something for a week if I can do it in one evening, with a negligible performance penalty.02:01
xiangfukristianpaul: I will send the package tomorrow. EMS was on holidays. 02:01
kristianpaulxiangfu: sure no rush ;)02:02
kristianpaulstill need to write the forth part..02:02
wpwrakxiangfu: so you decided to adopt azonenberg's design ?02:02
xiangfuwpwrak: yes. basicly modify from azonenberg's board. remove the ft232 chip. get the serial pins out for blue-tooth serial modular.02:02
xiangfuwpwrak: yes. 02:02
kristianpaulah good !02:02
wpwrakhellekin: you didn't read what i wrote after i got back home :)02:02
kristianpaulso getting back to qfp?02:02
hellekinI might have missed it :)02:02
xiangfukristianpaul: no.02:02
kristianpaulahmm02:03
Action: kristianpaul got lost02:03
xiangfukristianpaul: just another experience. I have plan to switch to fab-made-pcb + bga. 02:03
whitequarkwpwrak: as a trivial example, my flash decompiler is 8KLOC. eight thousand. it beats every single commercial one out there. yeah it works for five minutes on a single real-world file, but no one cares02:03
kristianpaulthat was earlier experience no?02:03
xiangfuearlier experience was the pre-made pin out pcb. only pins out. 02:05
kristianpaulso if i got right the diy experience is mostly done and what remain is out source manufacturing the boards?02:06
wpwraki'd avoid bga. this is not a good moment for it. bga has its place when you do serious quantities or have space constraints. right now, we have none of that.02:06
wpwrakwithout bga, you have the option of selling fully assembled board or just the pcb. since all the parts are from digi-key anyway, it's not too hard to make virtual kits.02:07
wpwraka pcb can be shipped as mail. very cheap and still reasonably quick.02:08
whitequarkwpwrak: my ruby implementation is almost ready for generating first useful code, and it's something like 4KLOC together with dependencies. tell me you can do the same in a month of full-time work in C, and I'll switch everything I have to it.02:09
xiangfuwpwrak: the xilinx new chips are all BGA. no qfp.02:10
xiangfuwpwrak: so if woflgang switch slx9 to 7 seriles. I have to use bga. :)02:10
wpwrakwhitequark: i wonder why you think C is generally inefficient. especially things that deal with bytes and such, instead of strings, are extremely straightforward there02:11
whitequarkwpwrak: experience with large codebases in both kinds of languages?02:11
wpwrakxiangfu: yeah, 7 would break all this. but then, is this coming ?02:11
xiangfukristianpaul: (source manufacturing the boards) no. those experience make me can make demo board fast, under controlled. :)02:12
whitequarkwpwrak: bytes, mostly, yes. but here is a project with a good amount of quite complex routing logic. it really doesn't go well with C (compared with more dynamic counterparts).02:12
wpwrakwhitequark: not really. lots of BASIC, Pascal, C. done some assembler, shell, Perl, Python (and of course Cora :). but for really big things, only C.02:12
sagexwpwrak, you are awesome02:13
whitequarkwpwrak: out of that list, only Python and Perl are dynamic HLLs, and it is generally hard to write big things in Perl for unrelated reasons02:13
wpwrak(plus the usual fringe languages, lisp, cobol, and all that. but that's just extremely superficial knowledge)02:14
wpwraksagex: thanks ;-)02:14
wpwrakwhitequark: when a project outgrows perl, i switch it to C :)02:14
whitequarkwpwrak: doh. sometimes I'll just find a few free weeks and show you how things should be done, with my code :)02:15
wpwrakkristianpaul: what i don't know is what wolfgang is planning to do about the board situation, if anything02:15
xiangfukristianpaul: the board i sent to you only have 1 pins out. 1 led, 1 DONE LED. the jtag pins name at back.02:16
kristianpaulwpwrak: me either... lets he speak by it self soon :-)02:16
wpwrakwhitequark: you know the risk in trying to teach that old dog is that he may just enjoy the meal (-:C02:16
kristianpaulxiangfu: well i do not plan more than a blinking led with the j1 cpu02:16
xiangfukristianpaul: ok. so you are using the Verlog code? right?02:17
wpwrakkristianpaul: so i looked around a bit, and for LX9, the situation is relatively bleak. the only thing halfway decent i found is that avnet board. (which is nice but has issues)02:17
wpwrakthere's also the question on whether we still consider such boards to be disposable (because experiments may destroy fpga internals)02:18
whitequarkwpwrak: well, I don't really care what you personally like, and I shouldn't. I think that if you'll see the actual benefit, you'll jump across yourself.02:18
wpwrakdisposable = board has to be very simple or modular02:18
whitequarkwpwrak: but have to finish this ruby thing first.02:18
wpwraknon-disposable = more convenient logistics02:18
xiangfukristianpaul: the UCF file I am using: https://github.com/xiangfu/mini-slx9-board/blob/master/firmware/blinking.xc6slx9-2-ftg256.ucf02:20
wpwrakwhitequark: (jump) i feel like a fish in water with C :) and i know enough of object-oriented principles and their application to C that i doubt it'll ever outgrow my needs. maybe my discipline may fail me from time to time, but that can happen in any language.02:20
kristianpaulxiangfu: yes all here https://github.com/kristianpaul/j1soc02:21
kristianpaulxiangfu: and yes i steal all your ucf and makefiles :-)02:21
xiangfukristianpaul: I can do quict test here too. in case you want more feedback. once you finish the j1soc.02:21
xiangfukristianpaul: great.02:21
kristianpaulxiangfu: ah yes, well no forth code yet..02:22
kristianpaulbut i will ping you when have dont that part02:22
kristianpaulalso will need load that in bram02:22
wpwrakanother option would be to move from LX9 to something else. there are also relatively cheap boards for non-LX9 chips. the question is how hard or easy it would be to adapt wolfgang's work to them. from what i've heard about what he's doing, this should be quite feasible, but may require some serious amount of grunt work.02:22
whitequarkwpwrak: sometimes forks take over the original projects, you know ;)02:22
wpwrakwhitequark: yet most die within a week ;-)02:23
whitequarkwpwrak: a week is optimistic :D02:23
xiangfu(but may require some serious amount of grunt work) yes. and he have to understand the chip details again. (just like wolfgang done with slx9)02:23
whitequarkwpwrak: looking at github stats, most don't even see a single commit02:23
wpwrakwhitequark: "within" is "<=" ;-)02:24
whitequarkyeah yeah :)02:24
wpwrakwhitequark: so it also includes the ones that already end after sobering up :) but then, the zero-commit ones are a bit suspicious02:24
whitequarkwpwrak: well, it's just that on github you're one click away from a fork. it is much more like a git-clone.02:25
whitequarkso you gotta have a lot of forks which aren't forks in traditional OSS sense. for example, the typical workflow is fork-commit-pullrequest02:26
wpwrakwhitequark: yeah, some may just fall asleep on their mouse :)02:26
whitequarkwpwrak: I think it is a side effect of github being a "social network" whatever that means. Some people mistake "fork" for "like" or something like that.02:27
wpwrak;-))02:27
whitequarkI mean that seriously btw02:27
wpwrakthe facbook culture :)02:27
xiangfukristianpaul: you are using the tqg144. the board I send to you are bga. the different is the clk pin and led pin are different.02:27
whitequarkwpwrak: kind of02:28
xiangfus/the board I send to you are bga/the board I send to you is bga/02:28
kristianpaulahm ok02:29
kristianpaulwell easy change ! :-)02:29
wpwrakxiangfu: so the idea is to produce one of these boards for fpgatools users ?02:29
wpwrak"users" being pioneers, of course :)02:30
xiangfuwpwrak: not sure. if someone want buy. we can sale. but the target is make some product for consumer. make some real money. 02:31
xiangfukristianpaul: yes.02:31
wpwrakheh :)02:32
wpwrakxiangfu: so what would the customer product be like ? fpgatools seem to be very very far from that 02:33
Action: kristianpaul thing same02:33
kristianpaulthink*02:33
xiangfuwolfgang already target on j1 soc.02:33
wpwrakin fact, i've been wondering if wolfgang had given up for good on making money :)02:33
kristianpaullol02:33
xiangfuwpwrak: I don't know.02:34
wpwrakhmm, what would be the consumer demand for a j1 soc ?02:34
whitequarkis J1 SOC the milkymist soc?02:35
whitequarksynthesized in asic?02:35
kristianpaulno no02:38
kristianpaulno no no :p02:38
kristianpaulwpwrak: is just a small target02:38
kristianpauli dont see any consumer demand eithe02:38
whitequarkkristianpaul: what it is then?02:38
kristianpaulbut fpgatools need do more that a counter anyway02:38
xiangfuwhitequark: http://excamera.com/sphinx/fpga-j1.html02:40
whitequarkahh, the forth critter02:41
whitequarkit is used as a simple yet real target for fpgatools, right?02:41
xiangfu(consumer demand) depends on what we want the product do with j1 soc inside.02:43
wpwrakxiangfu: well ... are there plans beyond J! ? as far as i know, J1 was the next step towards making a better "proof of concept"02:50
wpwrakwhat i don't see in all this is something sellable to end customers02:50
xiangfuno plan.02:50
wpwrakon the other hand, it's something that ought to be very suitable for attracting sponsorship from interested parties02:50
kristianpaulwpwrak: j1 or anyother small cpu can do actually something i think02:50
xiangfu( to end customers) I am not very clear on this. have to talk with wolfgang.02:50
jianjuDo you guys know of that Bittware board with an FPGA alongside an Epiphany chip?02:51
wpwrakusually, when there's some cartel in charge of things, there are many powerful entities who are silently suffering but will be more than happy to exercise their muscle as soon as they see a way out of their misery02:51
wpwrakkristianpaul: oh, sure j1 can do "something". but then, i can get a chip that does quite a few "something"s for 1/10 of the price of that fpga :)02:52
kristianpaulwpwrak: ;)02:53
wpwrakkristianpaul: now the question is what all this is supposed to connect to. maybe it's just wolfgang needing to work off some stress and there's no goal besides breaking the fpga conspiracy.02:54
kristianpaulbreaking fpga conspiracy02:54
kristianpaul:-)02:54
kristianpauljeje02:54
wpwrakkristianpaul: or maybe he's got some commercialization plan but isn't telling anyone02:54
wpwrakkristianpaul: or something else :) it's rather obscure02:55
kristianpaulhaha02:55
kristianpaulmaybe indeed02:55
wpwrakkristianpaul: it may also be that he doesn't see a way to turn this into income, yet there is.02:55
kristianpaulbut is too early?02:57
kristianpaulwell you need a plan to build on it02:57
wpwraknot sure about timing. depends on the goals. if there' s specific goal but it's still far out, perhaps keeping a low profile is good.02:58
kristianpaulyeah02:58
sagexoh guys whats happening02:58
kristianpauland you just announce fpgatools may get public officially ;)02:58
wpwrakbut then, if it's just to work off the frustration of fpga closedness, it would be beneficial to "go public" soon02:59
wpwrakand i think a blinking led is convincing proof of concept02:59
wpwrakkristianpaul: hmm, did you use "you" as "someone" or as "Werner" ?03:00
wpwrak(obviously, i hope there will be some sort of announcement in the fpgatools future :)03:00
wpwrak(and i do keep on putting hints into my monthly statistics. they're also reminders for wolfgang that he could be a little less secretive :)03:01
kristianpaulwpwrak: werner :)03:05
kristianpaulreminders oh yes i noticed most of then03:06
kristianpaulhehe03:06
sagexanyone use their nanonote to do something bizarre 03:08
wpwrakkristianpaul: good. so they weren't imperceptibly subtle :)03:08
wpwraksagex: ah, welcome to the nanonote club then :)03:09
wpwraksagex: of course, bizarreness is in the eye of the beholder. would this count for you ? http://downloads.qi-hardware.com/people/werner/ubb/vga/web/03:10
wpwrak(at least i tried to find a reasonably bizarrely-looking angle when taking the pictures :)03:10
whitequarkwpwrak: I just stumbled upon a problem and invented a solution for it05:57
whitequarkthen someone else realized I was going to rewrite `make'05:57
hozerdid they stop you?05:57
whitequarkhozer: I've to support Windows, and my data format is not serializable05:57
whitequarkbut at least I'd have a thing to steal code from05:58
hozerso are you installying cygwin and make now?05:58
whitequarkhozer: rewriting the part of make I need :305:58
whitequarkactually using make would be a terrible idea, as the amount of interfacing required to use that tiny part of the old tool will probably be far higher than the algorithm itself06:00
whitequarkbut it was a fun discovery nevertheless06:00
whitequarkI'm even solving basically the same problem: compiling translation units while obeying dependencies06:00
hozerheh06:01
hozerseems like we need a new title: code Archeologist06:01
whitequarkand he does what?06:01
hozeror engineering archeologist06:02
hozerfind a problem, look for old code that solves it06:02
whitequarkthis should be a prerequisite to be named an engineer even jokingly06:02
whitequarkthe amount of real (and imaginary) problems not yet solved by anyone publicly is vanishingly small...06:03
hozerit should be, but if that were a requirement most universities would cease to exist as they are06:04
whitequarkmaybe they should, then?06:05
hozerthe solution to renewable energy storage is sitting in 25-30 year old ammonia tanks06:05
whitequarkhm.06:05
whitequarkcould you elaborate?06:05
hozerYet DOE continues to talk about 'advances in fundamental blah blah blah for hydrogen storage'06:05
whitequarkalso that sounds quite... smelly. and not exactly safe in case of a rupture.06:06
hozerThere's more hydrogen in a truck loaded with pressurized liquid ammonia than a truck full of liquid hydrogen06:06
whitequarkhow come such obvious thing isn't mentioned anywhere?06:06
hozerwell, the big tanks are atmospheric pressure but -33f06:07
hozersmaller ones are pressurized .. http://www.extension.umn.edu/distribution/cropsystems/dc2326.html06:08
hozerhttp://7el.us/Solar/Thermo.pdf or http://freedomfertilizer.com06:10
hozerso basically, instead of making ammonia the way we've done it since the 1920's when Haber and Bosch figured it out (by using natural gas or coal for the hydrogen), we just use electrolysis to get the hydrogen, and the same haber-bosch reactor06:11
whitequarkhozer: this is an awesome idea. are there any other reasons (except for the ammonia being quite dangerous to handle) everyone is pursuing hydrogen stuff?06:12
hozerhydrogen is a pain to store06:12
hozerit either needs 5000psi (dangerous), or near absolute zero06:13
hozerand then, even as liquid, it's very low density06:13
hozerso about the only place the mass-energy-density of H2 makes sense is a rocket06:13
hozerwhitequark: where are you located06:14
whitequarkhozer: Russia06:15
hozerhttp://www.protonchemie.com/NFUEL.html is not too far from you06:16
hozerwell, I can't say that for sure, Russia is big ;)06:16
whitequarkMoscow to be more precise :)06:17
hozerah, proton did an ammonia handling terminal in Estonia06:17
hozerhttp://en.wikipedia.org/wiki/TogliattiAzot06:19
whitequarkhm, interesting06:21
whitequarkthanks for telling this, I never heard about ammonia storage before06:22
whitequark(as a hydrogen repository, that is)06:22
hozeryou can burn ammonia in engines or tubines directly too06:24
whitequarkI'd worry about releasing ammonia into atmosphere accidentally06:24
whitequarkthrough I'm not sure what would be the consequences of doing that with hydrogen06:24
hozerbut for places like Iowa that grow a lot of corn (which needs a lot of nitrogen), 20GW of wind turbines would make enough electricity to make the ammonia06:25
hozerdo you have cats?06:25
whitequarknot at the moment06:25
hozerwell, if you clean the catbox you have an ammonia release ;)06:25
whitequarkyeah, trace amounts of it. urea hydrolizes very slowly.06:26
hozerthe nice thing about NH3 is it's self-alarming. You know very quickly if there's a leak06:26
whitequarkwhereas here you could trivially get a large-scale chemical accident if your turbine malfunctions06:26
whitequarkhm06:26
whitequarkwhat if there isn't a small leak? on the page you've linked me to (    syntax do |s|06:27
whitequark      s.operand :tuple06:27
whitequark    end06:27
whitequarkerrrr06:27
whitequarkthis one: http://www.extension.umn.edu/distribution/cropsystems/dc2326.html06:27
whitequarkit lists quite a lot of hazards. failure of a safety valve, for example.06:27
whitequarkthat could quickly release a lot of ammonia without much warning signs06:28
whitequarkI don't want to be anywhere near that tank in that case.06:28
hozeryes06:28
hozerhttp://www.youtube.com/watch?v=sNkdAs1e7Cw06:28
hozernow here's the thing... what happens if you have a natural gas leak in your home06:29
hozerI used to haul ammonia tanks when I was in high school ... my dad had (well still has) an applicator06:30
hozerand then I decided I never wanted to deal with that stuff again06:30
hozerand 15 years later I'm doing a feasibility study on wind energy to ammonia06:30
whitequarkhozer: I have an electric stove here. I don't like natural gas either.06:31
whitequarkAlso: here, almost all length of the pipes is routed outside so that all leaks just spill to atmosphere, as opposed to accumulating and exploding in the basement06:31
whitequarknatural gas isn't dangerous unless it explodes, ammonia is06:32
hozerI wouldn't advocate putting NH3 in a home06:34
whitequarkI'd rather not see it in a city either06:34
whitequarkbut it is probably fine for long-distance transport06:34
hozeror long-term storage06:35
whitequarkyes06:35
hozerin the areas where it's windy06:35
hozerwe have all kinds of diesel fuel and gasoline in cities that just gets carcinogens in the groundwater :P06:35
whitequarkwe also have phenolic resins which release them right into our lungs06:36
whitequarkcarcinogens is one thing, caustic NH3 is another06:36
whitequark*are06:36
hozerwell, you can smell NH3 and easily avoid it :P, and then it evaporates and blows away06:37
hozerbut yeah ;)06:37
hozerand now, I need to get to bed06:38
whitequarksee you later06:39
hozerlater!06:40
whitequarkha, only two more years till mp3 patents expire07:01
wpwrakdo we need an  xclock --patent=mp3  that shows a count-down until expiration ? :)08:31
lindi-xclock doesn't support clipboard :(08:34
wpwrakwhy would we need that ?08:36
wpwrakand of course, that sounds like a nice thing for a twisted mind to add :)08:36
lindi-wpwrak: to paste the current time to some application?08:48
whitequarklindi-: alias whatever='date|xsel -b08:49
lindi-wpwrak: but that's the primary selection08:50
whitequarkwpwrak: I'm tired of installing libsox-fmt-mp3 and friends on every new linux system08:50
whitequarklindi-: no, keyboard08:50
whitequarkerr08:50
whitequarkclipboard08:50
whitequarkgotta get some sleep.08:50
lindi-ah true, but it leaves an extra process running :)08:50
lindi-Thu Jan  3 10:50:29 EET 201308:50
whitequarkleaves?08:50
lindi-yeah08:51
whitequarkhow so?08:51
lindi-"lindi    25218  0.0  0.0  15536   472 ?        S    10:50   0:00 xsel -b"08:51
lindi-now it's there sleeping08:51
whitequarkoh true08:51
whitequarkso, you've wasted 472K of system resources. what a pity.08:51
whitequarkoh and also one PID08:51
kyakwpwrak: when you talked about measuring of timer jitter, did you mean gettimeofday()?09:32
qi-bot[commit] kyak: libpurple: fix missing iconv (master) http://qi-hw.com/p/openwrt-packages/4112aca09:45
xiangfukyak: thanks on fixing packages.10:59
qi-bot[commit] Xiangfu: directvnc: fix depends +libjped (master) http://qi-hw.com/p/openwrt-packages/38c294710:59
kyakxiangfu: np.. trying to get it up and running11:00
kyakhttps://dev.openwrt.org/changeset/33909/trunk/package/libs/ncurses/Makefile11:14
kyakseems like a trivial change, but now for some reason --without-cxx-binding leaks into configure args, even though we have CONFIG_INSTALL_LIBSTDCPP=y in config11:14
kyakxiangfu: do you have an idea why it is happening?11:15
kyakthis leads to breakage of all ncurses apps written in C++ :)11:15
kyaki have a feeling that it must read ifneq ($(CONFIG_INSTALL_LIBSTDCPP),y)11:19
kyakyeah. .that's exactly how it should read11:24
kyaki'm don't want to start bashing OpenWrt again, but it's just not getting enough testing (and quality control) on some aspects11:25
kyakthis commit is almost three months, and noone even noticed.. it's a pity Ben is the only (more or less used) desktop target of OpenWrt11:26
kyaki'll post a bug report, just to clear my conscience and also see how this report is ignored11:28
qi-bot[commit] kyak: centerim5: update to latest git and fix missing iconv (master) http://qi-hw.com/p/openwrt-packages/1b8144611:42
qi-bot[commit] kyak: mpg123: fix missing deps (master) http://qi-hw.com/p/openwrt-packages/2b320f712:33
qi-bot[commit] kyak: jfbterm: fix dl url (master) http://qi-hw.com/p/openwrt-packages/88f3c3212:50
wpwrakkyak: yes, gettimeofday is what i was thinking of13:26
wpwrakwolfspraul: heya ! happy new year ! hangover successfully evicted ? ;-)15:32
wolfspraulhe, thanks! no alcohol no hangover :-)15:34
wpwrakooh, boring :)15:35
wpwrakwolfspraul: yesterday, we had a discussion here, trying to figure out where you might be planning to take fpgatools15:37
wolfspraulI'm not into these types of talks much15:38
wpwrakespecially regarding boards for experimentation (which will be needed in one form or another once people want to get involved)15:38
wolfspraulthere is a todo section in the readme, that roughly describes my current thinking15:38
wpwrakthat explains some of the core bits, but not much of the environment15:40
wpwrakwell, assuming you actually want to have one :)15:41
wpwrakfor example, if people want to "play" with your work ("play" as in low-key exploratory projects), what hardwere would they use ? e.g., do you plan to produce and sell xiangfu's bga board ?15:43
wpwrakor make the design and hope someone picks it up ? or evaluate and recommend a 3rd party product ?15:44
wpwrakor not care at all and expect others to take care of such worldly issues ?15:46
portconsolehello15:47
portconsolewhat is this irc for?15:47
wolfspraulnothing much works right now, so there is no need for a board15:48
wpwraki thought you had the LED blinking ?15:48
wolfspraulI could make 20 and send out as freebies maybe, don't know15:48
wolfspraulyes, at a fixed frequency15:48
wolfspraulnext step will be to make this more variable15:48
wpwrakbtw, i think the to do list lacks characterization: propagation delays, fan-out limits, maybe current restrictions.15:49
kristianpaulpropagation delays with larget designs indeed15:49
wpwrakfixed frequency is already a major achievement15:50
wpwrakit's like receiving the first greetings from extraterrestrials. even if it's just a "hello earth", i don't think anyone would dismiss it for lacking eloquence15:52
wpwrakportconsole: this channel is mainly about things related the qi-hardware universe: http://en.qi-hardware.com/wiki/Main_Page15:54
wpwrakthe old scare tactics still work ;-)16:03
kyak"read the topic or gtfo" would have worked, too!16:19
kyakbut you did it in a more elegant way, yes :)16:19
wpwrakconveys that air of superiority much better :)16:20
kyakwpwrak: so.. i'm calling my function at periodic intervals using nanosleep(). I also measure my function execution time using gettimeofday() and subtract that time from the nanosleep() intervals. I gave up the idea of real-time for now, but at least i know when my function has overrun16:30
wpwrakfor periodic calls, you may want to use interval timers. that way, you don't need to worry about adjusting the sleep16:31
kyakinterval timers?16:32
wpwraksetitimer16:32
wpwrakat least that's how i think they work. only used them once a long time ago, and that was for something else16:33
wpwraklemme check ...16:33
kyak"After the specified timer expires it generates SIGALRM which can be captured using signal() function to call a signal handler function."16:35
kyakperhaps it is what i need16:35
kyakheh, it's good. Even the code mimics the behavior of interrupts (empty while(1) loop)16:36
wpwrakyeah :)16:37
kyakthanks!16:37
wpwraknp :)16:37
kyakdamn, there are gazillion ways to do basically the same thing16:38
wpwrakthe joy of RT ;-)16:39
Action: wpwrak just realized something :)17:26
wolfspraulwhich is? :-)17:29
wpwrakthe ben may make a rather nice "environment" for an lx9 board17:30
wpwrakit can supply 3V3, a clock, and JTAG17:30
wpwrakso you just need a regulator (or maybe two ... dunno if you can get rid of the 2.5 V rail), caps, and some headers, maybe some leds17:31
wpwrakyeah, seems that Vccaux is fine with 3.3 V17:40
wpwrakoh, and there's even a lx4. is that new ?17:41
kristianpaullx4 dont think so17:55
kristianpauldamn 34 C  :-\17:57
wpwrak25 C here. very nice :)17:59
kristianpaul:-(17:59
kristianpaultemperature would nice if were some wind, but dont even that..18:01
kristianpaulwould be*18:01
wpwrakbut with a lot of humidity ?18:01
kristianpaulno humidy i'm at 1km high18:02
wpwrakno clouds up there ? :)18:02
kristianpaulbut its getting cloudy lets if it gets better18:02
kristianpaulyeap18:02
kristianpaulbut when is sunny there is wind18:03
wpwrakgot a nice view from the heights ?18:07
kristianpauli love my view from kitchen to the sunset18:07
kristianpauland from both batroom and kiving room to the sunrise18:08
kristianpaulthis is from kitchen http://kristianpaul.org/gallery/old_gallery_2/slide_05.html18:09
wolfspraulwe can consider the lx4 eol18:09
wolfspraulit was the same die as lx9 but with some things like mcb either disabled or maybe from lx9 yield18:10
kristianpaulwpwrak: not exactly living room but close http://kristianpaul.org/gallery/old_gallery_0/slide_01.html18:10
kristianpaulah yes missing mcb18:10
wolfspraulthe large and ongoing price reductions of the lx9 have taken over the small economic niche of the lx418:10
wolfspraultoday the situation is that nobody stocks the lx4, lx4 must be ordered in large quantities from xilinx, and price is equal or even above that of lx918:11
kristianpauli a shame there is no diy friendle packing in xilinx 7...18:11
wolfspraulif anybody still insists on lx4, I guess nothing can help them :-)18:11
kristianpaulis*18:11
wolfspraulI don't think much extra money is to be made if xilinx were to offer the xc7 in additional packages18:12
wpwrakwolfspraul: ah, digi-key have the lx4 and it's USD 10.9 (MOQ 1) vs. USD 15.7 for the lx9. but maybe they''re just dumping their stock.18:13
wpwrakkristianpaul: ah, nice view18:14
wolfspraulthe lx9 costs ca. 7 usd on the street in china now18:14
wolfspraullx4 needs to be preordered in large quantities at higher prices... in other words - noone buys it18:14
kristianpaulwpwrak: and it gets better on sunsets18:15
wpwrakwolfspraul: USD 7 is nice. it's a bit irritating that digi-key have no volume discount at all on those chips18:17
wolfspraulmaybe it's too easy to buy this from china? I don't know...18:18
wolfspraulchina accounts for >50% of global semiconductor sales nowadays, maybe for xilinx even more?18:18
kristianpaulbut buy from china online seems to lamost imposible if you dont read chinesse 18:19
wolfspraulyou just need to find one random china-based xilinx reseller and you will not care whether digikey has a volume discount for these chips or not18:19
kristianpauland write in english, i had have my bad experiences..18:19
wolfsprauloh sure, you have to find that one you trust and can work with18:19
wolfspraulbut in general I find digikey not the best source for 'key' chips18:19
wolfspraulmaybe because they are 'key', so customers tend to secure supply for these manually/individually anyway18:20
wolfsprauldigikey is good if you need high-quality and reliable source for all the other small & important stuff in your designs18:20
wolfspraulbut for the 3-5 key chips, maybe not18:20
wolfspraulthey won't be there at all, or no volume discount, etc.18:21
wolfspraulthat's my experience at least18:21
wpwrakyeah, the fancier the thing, the worse the conditions18:21
hozerwolfspraul: do you have schematics/pcb for the lx9 board you are using for fpgatools?19:29
wolfspraulsure19:40
wolfspraulI have no local storage, quite simply19:40
wolfspraulsome are hand-assembled with standard testing pcbs though, and some are variants of azonenberg's board I think19:40
wolfspraulthere is nothing unique or valuable in those "designs"19:40
hozerI have a XESS Xula2 board with an LX25-256bga19:42
hozerand I've been thinking about how to get a generic small form factor board that could take lxN-csg324 chips19:42
kristianpaulhozer: hi, how did end your lm32 soc port btw?20:09
kyakwpwrak, kristianpaul you poor guys, i was just skating :)20:13
kristianpauldamn you lucky! :-D20:17
qi-bot[commit] kyak: qstardict: update to compile with latest glib2 (master) http://qi-hw.com/p/openwrt-packages/0267b8820:47
hozerkristianpaul: I got lm32 + milkymist to synthesis, but the board I have has a 16 bit dram instead of 32 bits so there are hanging data lines ;)20:50
qi-bot[commit] kyak: sdcv: fix build (master) http://qi-hw.com/p/openwrt-packages/730b03820:51
kyakopenwrt devs updated glib2 version. Now, atk won't build against this version (so again, that wasn't tested). If we start updating atk, it would lead to gtk2 update and that would automatically break gtk2 interface on Ben, since currently used version of gtk2 is the last one where directfb is working21:19
kristianpaulhum..21:22
wpwrakopenwrt, the eternal uphill battle ... well, you know my opinion. you (mainly kyak and xiangfu) are actually doing an admirable job, but oh the humanity ...21:23
kyakback in those days, you would mention OE and Jlime. Are you still convinced it would've been a better choice?21:25
wpwraki think with enough effort, OE/Jlime could have survived, yes. rafa's involvement was time-limited because he eventually had to get a job, but i don't think OE is any harder to maintain than openwrt. (i mean, neither is easy but each has its own quirks)21:27
hozeris OE debian-based?21:27
wpwrakhozer: unfortunately not21:27
hozerwhy not emdebian21:28
wpwrakwhich brings me to the next step: rafa also suggested that debian may be a choice. debian had many issues back then (with regard to embedded systems), but given general tendencies, i would have favoured something going towards debian21:28
kyakwe would see same problems with OE.. their repos have gtk+_2.24.14, where directfb is long broken21:30
kyakdunno.. of course we can override glib2 from openwrt and just run the older version21:31
kyakbut then we can take it further and run older everything :)21:32
wpwrakkyak: i think OE should be more open towards "desktop" environments. for openwrt, they're (understandably) quite alien21:32
kyakyes, this is true21:32
wpwraki also understand worfgang's aversion towards OE. openmoko was the nonstop pervert party of OE. people there used it as a fancy "make". and he was in charge of overseeing that madness.21:33
kyakso openmoko was OE-based?21:34
kristianpaulwpwrak: debian good old dog :-)21:35
kyaki didn't quite understand what do you mean, could you explain more?21:35
wpwrakso obviously, if you wake up from nightmares about OE, go over to meet you folks who have cooked up even scarier things, struggle the whole day to control the fallout of OE perversion, then maybe spend some time watching your kernel hackers fight, before you fall into bed, exhausted, and then the nightmares return, then i guess it's understandable that you don't choose OE for your next project :)21:36
wpwrakkyak: yes, openmoko was OE-based21:36
kyaki'm just trying to understand where is OE's fault exactly21:37
wpwrakkyak: and the OE was a big resource drain. even worse, there was no clear separation between distribution and regular development. so everybody in the sw group lived too close to OE21:37
wpwrakkyak: well, except for a few, myself included, who managed to "outsource" the OE nightmares :)21:38
wpwrakkyak: ah, it's the overall complexity21:38
wpwrakkyak: and, in the case of openmoko, that there are no clear separation between distribution and general development21:39
wpwrakkyak: so most people developed their pieces of code inside the OE environment (using bitbake instead of make, and so on)21:39
kyakin openwrt, base system is separated from packages by using "feeds", which contain those packages. Is it the separation you are talking about?21:41
wpwrakno. i actually don't quite get that "feeds" division.21:42
wpwrakin OM, the (well, one) problem was that people developing applications developed them _in_ OE21:43
wpwrakas opposed to the traditional approach where you do your thing and it you need something from the distribution or you and the distribution to pick up something, you'd talk to the distribution guys21:44
wpwrakthis division didn't exist. therefore, any OE hickup immediately became everyone's problem.21:44
wpwrakand also everybody had to understand OE.21:44
wpwrak(and install it, etc.)21:45
kyakyep.. i understand now21:46
hozerah, this explains a few things now21:51
hozerwhen I was a MontaVista we checked everything into Bitkeeper repos and used RPM packages21:52
hozerand then we had to migrate to CVS/SVN 'very quickly'21:53
hozerthere were a few things I liked about OE, but it was brittle21:53
hozerSo what if we cloned debian unstable, and had a 'Dem' (Debian EMbedded) (or emD, or something) that was like testing, and finally a stable branch21:55
hozerso in the case of Debian, stuff doesn't get promoted to stable until it's been in testing awhile.21:56
hozerI want to do an embedded system distro that puts things into 'stable' after they pass a regression test suite21:56
hozerwhich would hopefully block the kinds of nightmares that OE had21:57
kyakwell, there are releases in openwrt22:00
kyakbut i'd say that they happen very often, so we stick to a particular svn revision when releasing an image for Ben22:01
kyak*don't happen often22:01
hozerright, but it's hard to get GUI stuff pulled into openwrt, right?22:01
kyakGUI, i18n, UTF-8...22:02
kyakit's very sparsely tested, if tested at all22:03
wpwrakat OM, our main reference was the local OE anyway. i don't know how rapidly things were sync'ed to main OE (probably quickly, since we had some mainline OE developers in our team)22:04
kyakhttps://dev.openwrt.org/ticket/12317 -> things like that don't get noticed, because noone in openwrt world is even using ncursesw22:05
kyakand these bugs go low priority22:05
kyakthis is at least my understanding22:06
kyakindeed, who cares about ncursesw, if there are lost packets on some router22:06
wpwrakwhat i did was that i maintained a border between the things i did and OE. when i wanted to put something into OE, i told the OE folks and they took care of the rest. and when i needed a package from them, i asked for it and it usually showed up soon thereafter. of course, my work wasn;t very intertwined with OE, so this made my model of operation more feasible.22:06
wpwrakborder, or boundary22:07
wpwrakkyak: i'm a bit surprised they wouldn't care about console things (like ncurses)22:07
kyakwpwrak: it's wide ncurses :)22:08
kyakthe variant which supports UTF-822:09
wpwrakwell, whatever-ncurses as long as it's not xncurses ;-)22:09
kyaki think even xiangfu gave up on submitting patches22:10
kyakthree months ago we made a release and had to use additional patches for qt422:11
kyakthey are not in openwrt22:11
wpwrakmmh, that doesn't sound good22:12
wpwraki thought owrt folks were quite close to qi-hw ?22:12
kyakyeah, i'm so happy when something is broken and i can fix it without interferring to upstream openwrt22:12
wpwrak;-)22:13
kyakthey are right in this channel, yes :)22:13
wpwrakkinda absent, though22:15
sagexhmm anyone have a successful and stable usb-ethernet solution for the nanonote22:23
kyaksagex: i do, following the instructions in "ethernet over usb" article in wiki22:25
wpwraksagex: mine looks like this: http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/bin/ben22:26
sagexthats what I was looking for22:26
sagexthanks kyak & wpwrak 22:26
wpwraki noticed that it can take a few tries with recent openwrt. not sure why. so don't give up if it fails at first22:27
sagexYeah what network interface do you have on your host machine22:28
sagexI run the default kubuntu network interface22:28
--- Fri Jan 4 201300:00

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