#qi-hardware IRC log for Monday, 2010-10-18

wolfspraulwpwrak: (you are probably sleeping, but I just stack it here...) the component references (last field) in the order list (.ord file) are the references from the original kicad bom, correct?00:21
nitin_guptaneed some help with micro SD card01:17
nitin_guptacan anybody help01:17
wolfspraulwpwrak: hmm. another advantage of using the .csv coming out of pcbnew is that I believe one .brd file always exactly equals one board, but multiple .sch pages can lead to one .brd02:45
wolfspraul(all questions actually)02:45
wolfspraulso thinking about automating some things on the server, we could auto-generate the .csv bom more safely for each .brd. for the .sch, we would need to find the root .sch like schhist?02:46
wolfspraulwpwrak: clarification for the parts list documentation (.par), README 240ff. does each reference have to appear only once in the file, or will multiple lines refering to the same reference be merged?02:55
nitin_guptai want to suppress error message in linux script02:59
nitin_guptahow to do that02:59
kyaknitin_gupta: find out where this error message goes from and solve the root of problem03:07
qi-bot[commit] Xiangfu Liu: [uboot-xburst] small cleanup about __gpio_jtag_to_uart0() http://qi-hw.com/p/openwrt-xburst/d1f331b03:08
kyakotherwise, just > /dev/null 2>&103:08
wolfspraulwpwrak: why is the description file .dsc not included in the inventory file .inv?03:23
nitin_guptawhile doing ssh getting "permission denied , publickey gussapi- with -mic password"03:23
nitin_guptaI am not able to login into NN using ssh or scp03:23
wolfspraulwpwrak: in CHARACTERISTICS, line 63, it says "Each catalog entry begins with a part number followed by a part type designator."04:10
wolfspraulI assume the CHARACTERISTICS file is a bit older, and meant to describe the .chr file mentioned in README?04:10
wolfspraulif so, should this be "Each catalog entry begins with a namespace followed by a part-number."?04:11
wolfsprauloh maybe it's namespace followed by part-number followed by part-type-designator? not really sure whether the designator is a must, also CHARACTERISTICS doesn't seem to list all designators04:13
qi-bot[commit] Xiangfu Liu: move mmc files to drivers/mmc/ http://qi-hw.com/p/openwrt-xburst/d2f389b04:18
wpwrakwolfspraul: (.par) in the file, each reference for which the system found a suitable entry appears exactly once. (it actually does the full lookup for each, too. there's some potential for optimization)04:20
wpwrakwolfspraul: (.csv vs .lst) yes, you'd have to find the root for eeschema. i don't know if both formats are equivalent, though.04:21
wpwrakwolfspraul: (.ord) yes, these are R123, R124, etc. one example from atusd.ord: DIGI-KEY RMCF0402ZT0R00CT-ND 2 USD 0.04 R1 R204:24
wolfspraulhey :-) now I need to check back in the logs what my questions were :-)04:24
wpwrak(scrolling back .. ah, there's more :)04:24
wolfspraulI just had another one - do you foresee that there will be a global .sub and .gen, or can they be project specific?04:25
wpwrak(.lst documented) i kinda doubt it ;-)04:25
wpwrak.gen is specific to the manufacturer and maybe even for parts families. maybe some manufacturers can share a gen but that would be somewhat unusual04:26
wolfspraulfor .csv vs. lst, after understanding sub and gen I think you will want the flexibility in the custom fields, F1..F8, FN, etc. I believe those things are missing in the .csv generated by pcbnew04:26
wpwrak(maybe if you express something like 74xx via characteristics, you could share it. haven't really thought of this. it's probably very tricky.)04:27
wpwraka .sub could be shared by all the projects that use a common annotation style04:27
wolfspraulbut it looks like there is nothing project-specific in .sub or .gen04:28
wolfspraulat least we could say so and make the system more powerful, of course the projects on the qi-hardware server then have to follow that usage of the fields04:28
wolfspraulwouldn't it make more sense to accumulate site-wide .gen and .sub over time?04:28
wolfspraulI see it more as part of boom than as part of a project.04:29
wpwrakthe idea is that .gen can be shared among projects. e.g., all projects can use the same murata parts characteristics. much like everyone uses the same parametric search at digi-key.04:29
wolfspraulbut you tell me...04:29
wolfspraulyes exactly04:29
wolfspraulsounds like we should just make site-wide (qi-wide) defaults for those things, I don't see why anybody would want to override them right now since nobody uses the system yet anyway04:30
wpwrak.sub is more subtle. i would think of some sort of style guide that explains exactly how to annotate things. maybe you remember that we spend quite some time in gta02-core to define these details. e.g., 1.2 kOhm vs. 1K2 vs. whatever04:30
wolfspraulyes of course04:31
wolfspraulI followed closely because I think it's quite important.04:31
wolfspraulpeople can do whatever they want, but I am wondering what we should automate on the qi server04:31
wolfspraulwe can just declare a default notation with which boom works best, no?04:31
wpwrakboom is agnostic to what you to in .sub. it's even fairly agnostic to how you specify things in .chr. it's the .sub that brings the two together.04:31
wolfspraulI mean 'default' for Qi04:31
wolfspraulyes I understand, but I'm thinking about running this on the server04:32
wpwrak(default qi style) yes04:32
wolfspraulthe notation you defined for gta02-core is perfect for me04:32
wolfspraulI mean I have zero arguments why we shouldn't just use it everywhere.04:32
wpwrakbtw, there can be more than one .sub file :) a project may have some "local" quirks. or explore areas the shared policy hasn't touched yet.04:33
wolfspraulit's not about forcing someone into a notation, it's just that maybe at this time, I don't need to think how we can override project-specific .subs when nobody is using them anyway04:33
wolfspraulbut it sounds like we can start with a qi/site-wide .sub and .gen system04:33
wolfspraulbecause the data in these files either has nothing project-specific (.gen), or should not if they follow a well thought through notation (.sub)04:34
wolfspraulwell whatever you tell me, I'm just reading and thinking04:34
wpwrak(reuse gta02-core style) depends a bit. the gta02-core style was also influenced by gta02. i tend to prefer things a little more verbose. i also find adam's? carlos'? compact way of adding tolerances interesting.04:34
wolfspraulI think I worked my way through the old svn boom, I continue in the git boom now :-)04:34
wolfspraulyes but we are small now, we can just come up with one notation and make it the default.04:35
wolfspraulI'm not against overrides but it soudns like right now (!) they are not needed, neither for .gen nor for .sub04:35
wpwrakfor .chr, you may also have new parts. i like the ability to "test drive" things in a project and to only push them into the common repository once they're sufficiently validated.04:36
wolfsprauldid you see my comment about CHARACTERISTICS line 63/6404:36
wpwrake.g., i do the same with footprints. i keep them in the projects until i'm happy with them. only then may they migrate into kicad-libs.04:37
wolfspraulof course, makes sense04:37
wpwrak(more questions) getting to them ... LIFO order ;-)04:37
wolfspraulcorrect me if I'm wrong, my impression is that bom2part tries to deal with the technical/electrical aspects, while part2order tries to deal with the business/financial aspects04:37
wolfspraulis that a fair characterization of the design?04:38
wpwrakwell, random order :)04:38
wolfspraulwpwrak: look at that :-) http://en.qi-hardware.com/wiki/Values_and_units_in_schematics04:40
wolfspraulwritten up Oct 27, 2009...04:40
wpwrak(comp by value) this one isn't used and it's simply ignored if it's there. if you check comp by value but omit comp by ref, then boom will get confused, because it finds them by position not by any name. (anything that could serve as a designator is cleverly i18n'ed, so there's no way of reliably identifying them. well, i could just parse it and see if one is always Rnnn followed by non Rnnn or such. ugly heuristics, though.)04:41
wpwrak(sub comp) not really sure what this does04:41
wolfspraul[comp by value] ok I will leave comp by value unchecked then, it's not meant to be checked in the context of boom. understood.04:43
wpwrak(CHARACTERISTICS) it's older and possibly inconsistent, like it says at the very beginning ;-) it really ought to become a style guide and not so much a file format description.04:45
wpwrak(CH...) it's actually  names-space part-number field=value tuple(s)04:47
wolfspraulyeah that's what I thought. so what is called 'type designator' in CHARACTERISTICS is just a regular field now04:47
wpwrak(CH...) the part type is no longer special. e.g., you could even have a hierarchy with sub-part-types.04:47
wpwrak(CH...) the part type is now expressed as T=part-type. and i also made the designators shorter, R, C, ... (see the .sub file :)04:48
wolfspraulwhich .sub file?04:49
wolfspraulI am still exclusively in the old svn boom04:49
wolfspraulI think I am ready for the next level, will move to the git boom now :-)04:49
wpwrakthe .chr file format may change even more. e.g., with a better parser, perhaps i could omit the field names and have a more table-like structure instead. the background is that .chr is something a human may occasionally have to write manually, and there, one would want to avoid having lines > 80 characters. so anything that helps keep things short is good.04:50
wpwrak(which .sub) gta02-core/bom/gta02-core.sub, ben-wpan/bom/atrf.sub, eda-tools/boom/test/test.sub, they're all very similar04:51
wolfspraulyes I'm looking at the test.sub now04:51
wolfspraulok I will move to git boom now and dig around there04:51
wpwrak(.dsc not included in .inv) mainly for parser convenience. BUT ... you could also have different sources of descriptions. and you may even switch among them in the course of processing.04:54
wpwrake.g., your EE design group could use descriptions conforming to some "local" standard, while the sourcing group may want to look at the supplier's descriptions (which may be of varying quality)04:55
wolfspraul[.dsc not in .inv] yes and I noticed in general you try to keep things in separate files a little, like relational tables in the filesystem. rather than loading things into fewer files. makes perfect sense I was just curious whether there were any special considerations.04:55
wpwrak(gta02 is a good example for when it makes sense to override descriptions. the fic descriptions are almost always more detailed than the ones by digi-key. also, if you have, say, chinese suppliers only providing chinese descriptions, or the same with spanish, then you may want to use an english version for a globally shared project)04:58
wpwrak(relational tables) exactly :) that's what all this really is in the end04:59
wpwrak(quantity field in pcbnew's .csv BOM) the idea is that this BOM would be worked on by human BOM processors. they wouldn't like to do the same part 100 times :)05:01
wolfspraul[quantity field] good point. although the number of comma separated items in the "Designator" field probably always must match the Quantity field. But sure, for humans looking at a number is far better than counting commas.05:02
wpwrak(qty) it also helps to keep the keys in the table (if you consider it as one) unique. boom shouldn't really care. and .par should have a quantity field.05:02
wolfspraulwell I think that decision (pcbnew's bom vs. eeschema's bom) is already settled05:02
wpwrakin fact, .par is one file i could imagine getting rid of05:02
wolfspraulyou need the custom fields, and afaik they are only in eeschema's bom05:03
wpwraki don't now yet. all i know is that eeschema's bom works for me :)05:03
wpwrakalso, the bom stuff in kicad my change before too long. there's been talk about this.05:04
wolfspraulwpwrak: yes it may change, but given the scope of boom I don't think kicad will attempt something so broad?05:07
wolfspraulwhich means that whatever changes will just trigger some 'smaller' changes for boom05:08
wolfspraulor no?05:08
wpwrakyes, i don't see any real issues there. just that it doens't make much sense to "optimize towards the current implementation"05:10
wpwrake.g., the absence of a dialog may not last05:10
wolfspraulwpwrak: would you agree with my characterization of the bom2part/part2order design?05:16
wpwrakhmm, which characterization of these two ?05:26
wolfspraulwpwrak: my impression is that bom2part tries to deal with the technical/electrical aspects, while part2order tries to deal with  the business/financial aspects05:28
wolfspraulhow can I help with boom now? in which way can it run on the server?05:29
wpwrak(tech vs. biz) yeah, that's pretty good05:33
wpwrak(run on the server) i think it's a bit premature for this. we first need some more infrastructure.05:34
wolfspraulok my impression is that bom2part tries to deal with the technical/electrical aspects, while part2order tries to deal with  the business/financial aspects05:34
wolfsprauloops :-)05:34
wolfspraulok I will try to use it locally a bit then05:35
wpwrak(help) part of the infrastructure would be more parts definitions, be it equivalences or characteristics :)05:35
wolfspraulalthough right now I'm a little confused because there is so much construction going on05:35
wolfspraulthere's the stuff in the svn.openmoko location, then the stuff in git/eda-tools05:35
wolfspraulseems no connection between the two?05:35
wpwrakalso, the characteristics need to get fleshed out a bit more. e.g., there my be relevant parameters i'm not including of that i'm not handling well05:35
wolfspraulyou talk about C, but I can't see that committed yet?05:35
wolfspraulwell that's the hardest part for me to help, lack of competence05:36
wpwrakrome wasn't built in a day :)05:36
wolfspraulI'm very eager to learn, but the best way to do so is probably by following/reading the boom sources, and trying it locally for some small projects05:36
wpwrakthe boom "core" is still in svn. the git part is mainly about building a better database for now05:36
wpwrakthere's not a single line of C so far. only in my head ;-)05:37
wolfspraulok next step for me: try to run it locally05:37
wpwrakyup :)05:37
wpwrakregarding schedule, i don't think boom will be ready for general consumption before the end of the year. it's now a t the point where it can do some nice tricks, but you still fall into a hole where you have to provide infrastructure very often.05:38
wolfspraulthat's all fine, but I believe the only way to get it over the mountain is to start using it in reality05:39
wpwrake.g., this weekend's burst of activity was basically just to help me buy a few dozen different resistors.05:39
wolfspraulso I will do my part on that, those discussions here help, and we take it from there05:39
wolfspraullike you said, Rome was not built in a day05:39
wolfspraulI don't expect to do an apt-get install boom and then somehow magically 'everything' works, where we are not even clear what that 'everything' is. no worries.05:40
wolfspraulyes I can imagine :-)05:40
wolfspraullike I said, need-driven, very good05:40
wolfspraulbut a goal of getting this to the point of 'general consumption' like you say, or that every KiCad project on the qi server can benefit from it, is a nice goal05:41
wpwrakthe problem with people using it now is that all these holes become blockers for them. and uncharted areas may need some improviding (lest they become blockers, too), so a legacy incompatible with future changes may grow there05:41
wpwrakthat's why i want to do a bit more trailblazing first05:41
wolfspraulfully agreed05:41
wpwrakright now, the best "use" of boom is to basically play with it. take an existing project, generate the bom, vary things a little. stuff where the basic are known to work and where you can always walk away from things that don't work05:42
wolfspraulso it's perfect for me :-)05:44
wpwrakputting it on the server is the least of my worries :) interactive use is more important if you need to make a real shopping list. the one on the server would likely be more informational.05:44
wolfspraulI can learn more about those pesky parameters and what you call 'characteristics'05:44
wolfspraulwell that depends, I can definitely imagine the server being very very useful05:44
wpwrak(or get complex. e.g., how would i include local suppliers ?)05:45
wolfspraulespecially if we collect large components databases on the server, and if we slowly improve the .sub and .gen files05:45
wolfspraulyes sure05:45
wolfspraulI would do this step by step05:45
wolfspraulfirst make it run, with global defaults05:45
wolfspraulpeople see the results, use them either as-is, or in copy/paste fashion, or they install boom locally05:45
wpwraki don't like the idea of a web-centric system where people maintain their files (supplier preferences, supplier-specific equivalences, etc.) in some central database05:46
wolfspraulyes I agree05:46
wolfspraulbut seeing the results with global defaults, and some strong server-side databases and .sub/.gen sets will make them want more05:46
wolfspraulI am in no way going to implement some account system with uploads of preferences etc.05:46
wolfspraulno time for that :-)05:47
wpwraka "one size fits all" web interface would have to be simple and would be somewhat limited in what it can provides. maybe you could have a few basic profiles to pick, e.g., "show me US or EU sources", but that would be the most i'd do there05:47
wolfspraulgiven our projects and people, most actual runs will be < 100 anyway05:47
wolfspraultotally agree05:47
wolfspraullike I said, the key is to show the power, show a good default05:47
wpwrakit's not the run size :) in fact, smaller runs are even more sensitive to things like shipping charges05:47
wolfsprauland to have nice server side databases with parts from the big manufacturers, distributors05:48
wolfspraulI don't even see it like that. I just want to make this bom/shopping experience less mysterious, get some computing power behind it.05:48
wpwrak(show the power) yeah. getting people to migrate to boom (instead of excel-based sourcing or whatever) is good.05:48
wpwraki hope cpu power won't be the bottleneck :) i'm more concerned about database size and update rate05:49
wolfspraulalso like I said, I see big value in large manufacturer/distributor databases on the server, and a slowly optimized global set of good .sub/.gen files05:49
wolfspraulif cpu power is a bottleneck, I'll get bigger servers :-)05:49
wolfspraulthat's the whole point, let the servers do some work, show it to people in a browser05:49
wolfspraulat that point they have many options05:49
wolfspraulno accounts, no upload of preferences, definitely no shopping or payment system routed through the qi servers05:50
wpwraki think you need a local copy of the database. going to the network for each query would be way too slow. running the db on the server would complicate the design.05:50
wpwrakwhat the server can do is simply answer the FAQ "how much are the parts ?"05:51
wolfspraulwhy not just run the whole thing on the server? fully automated05:51
wpwrakgive people some idea of what a project costs05:51
wolfspraulah yes, of course05:51
wolfspraullike I said, it should all be fully downloadable05:51
wolfspraulbut at any given time, the server can have actual large databases of manufacturers and distributors locally (on the server's hdd)05:51
wolfspraulso it can show what boom can do05:52
wolfspraulthat's not a bad thing, and not at all meant against downloading and running things locally05:52
wpwrak(run on the server) doesn't work too well for doing new things. also, given that you already have a large database, you can do more with it, e.g., do a parametric search to select component values.05:52
wolfspraulyou mean if you have that database locally?05:53
wolfspraulI fully agree with you.05:53
wolfspraulI am thinking about another case.05:53
wpwrakfast turn-around times. etc. :)05:53
wolfspraulsome newbie comes to our server05:53
wolfspraultotally from outside05:53
wolfspraulhe is thinking about taking a project, say xue, and turning it into something different for his customers05:54
wolfspraulby _showing_ him boom in action, live on the server, he can get a better feel for how much time & money he needs to invest from just browing through the project files, to being able to sell stuff to his customers05:54
wolfspraulthat's a big black hole right now05:54
wolfspraulof course _you_ can imagine, because you know boom already05:54
wolfspraulbut how do we visualize it?05:55
wolfspraulI think by just installing and running it on the server, with global defaults.05:55
wpwrakhmm, seeing boom in action won't tell him that much :)05:55
wolfspraulif you then go ahead and actually do a run (maybe even a customized/derived product), in many cases you will take everything local.05:55
wolfspraulsure he sees how easy it is to create shopping lists05:55
wolfspraulthat gives him confidence that the entire project is actually usable in real life05:55
wpwrakwhat seeing boom in action does tell you is what things will be like for sourcing05:56
wpwrak(well, a little. i currently don't show all the goodies. but that could change)05:56
wolfspraulyes. and that's a great value I think.05:56
wpwrakit will show him that a way to create a shopping list exists, correct05:56
wpwrakthen there will be the BIG style guide explaining what rules to follow :)05:57
wpwrakand then a download of a few MB (compressed) worth of database. install BOOM and you're set :)05:57
wolfspraulof course if you make a decision to join, produce the product or derived/customized product, you take things locally and it will still work05:58
wolfsprauland with real life data, real manufacturers, real distributors05:58
wolfspraulso when he decides to take on his project, he has more confidence that those things will work, and he can focus on his customizations05:58
wpwrakbtw, one of the reasons why i added a ton of common resistors and caps is to eliminate one such black hole. it's very annoying if even just changing one resistor value means that you have to switch to sourcing mode and play with digi-key for ten minutes to bring the BOM up to date again.05:59
wolfsprauloh sure05:59
wolfspraulI think we see the potential of boom similarly05:59
wolfspraulso: I will first play with it locally, and the goal is to get it to the point that it can automatically run for every KiCad project on the qi projects server05:59
wolfspraulin a few months...05:59
wolfspraulI finished my introductory reading session :-)06:00
wolfsprauland thanks a lot for answering all my questions!06:00
wpwrakas i said, i wouldn't worry much about the server for now. that will be a byproduct. the important use is local.06:00
wolfspraulthe server is our advertisement so that there will be more local use06:01
wpwrakand there are still a number of prerequisites before local use is ready for prime time. including rewriting the mess in Perl, and fattening the database06:01
wpwrakyes, it can work as advertizing06:02
wpwrakonce prerequisites are done, then we can work on the style guide. this will be a lot of work, also because there are many exotic components to deal with.06:02
wpwrakfor now, i just focus on very mainstream parts, but each project needs a few "strange" items06:03
wpwrakalso, i completely ignore packaging for now. so all you get are "cut tape" prices and order numbers. no tape and reel. you'd have to do that manually.06:03
wpwrak(it's not too horrible a task, considering that you're probably putting some serious work into it when you start thinking about buying t&r. but of course, if automation can save a day or two in digi-keys' online order system, that's welcome)06:05
wolfspraulall good06:05
wolfspraulstep by step06:05
wolfspraulfirst you have a second user now :-)06:05
wolfsprauland more clueless user, which is a good test...06:05
wpwrakyep :)06:05
wolfsprauloh and it's not just digikey06:05
wolfspraulif it would be a digikey only system I would be far less interested06:05
wpwraksure, you can add any distributor :) digi-key just have what's probably the nicest database06:06
wpwrakthen it goes all the way down until you finally reach those gazillion "call for a quote" shops all around asia06:07
wpwrak(and their sometimes quirky part numbers)06:07
wpwraki should probably also have a table that shows when a part's record was fetched from the distributor06:08
wpwrakparticularly if the process is manual, things can get very old06:09
wolfspraulthe system is flexible enough to accomodate more large-scale sources later06:09
wolfsprauland it is strong on characterizing parts (or will be strong :-)) dealing with substitions, equal parts, etc. etc.06:10
wolfspraulif it works well one day, it's a huge difference to make free designs a hardware reality for someone who wants to make money with them, not just look at them in admiration06:10
wolfspraulthere are serious sourcing companies in Asia too, of course. and boom will easily accomodate their 'api' later.06:10
wolfspraulgood idea06:11
wpwraksounds good, yes. the idea is that boom is very open. that's also why documenting a good workflow is important. boom will not constrain or guide you much, but the consequences might :)06:11
wolfsprauldocumentation you have now is not bad06:11
wolfspraulthe xfig chart is nice, although some fic parts in there are probably outdated06:12
wpwrakyeah, the chart needs work :) there's also new stuff that's not included06:12
wolfspraulthe order the files in README are described is a bit strange, it starts with the 'harder to understand' concepts like .sub and .gen06:12
wolfspraulI read it bottom-up06:12
wpwrakREADME is more or less in the processing sequence. but yes, bottom-up may work better.06:12
wpwraki'm not too unhappy with README. but that's the plumber's guide to boom06:13
wpwrakCHARACTERISTICS is what should eventually be the EE's manual06:13
wolfspraulit's nice06:14
wpwrakthe EE's manual should also discuss various styles. show how they can be implemented. explain a bit the advantages/drawbacks.06:15
wpwrakmaybe have a multiple choice for each style decision. so you can define your project's rules just by checking items :)06:16
wpwrakboom also needs more work on the error reporting. e.g., it's relatively easy to miss it if it doesn't find a part. there's no explicit "what went wrong" report, only what the individual stages print at run time.06:17
wpwrakand it debugging something that went wrong is difficult. particularly if there problem is in .sub (can also happen if you feed it bad input, so you don't have to be writing the .sub yourself)06:19
wpwrakso, still a bit more work. we'll get there, eventually :)06:19
kyakB_Lizzard: hi! how is your luck with nupdf?08:36
B_LizzardGave up :D08:37
B_LizzardFor now08:37
B_LizzardSome of the stuff needs a lot of work08:37
B_LizzardChanges between mupdf 0.5 and 0.6 are very large08:38
kyaki would say that nupdf requires for complete re-write, as it's underlying library, mupdf, was completely re-written08:38
kyakbut you could have more luck with mupdf X11 app itself08:39
kyakcause there is X11 in Jlime08:39
kyaki would like to give a try to fbida (which offers fbgs that can display pdf files with the help of ghostscript)08:40
kyakthis, of course, requires gs and fbida :)08:40
B_LizzardWell, from my experience with the HP Jornada, the only things that will work is mupdf and xpdf, in a lesser degree08:41
B_LizzardSome of the mupdf keybindings don't help but that can be fixed08:41
B_LizzardHaving an SDL frontend is nice but if it doesn't compile it's useless :D08:42
kyaksame to be said about the mupdf lib.. it might be nice and fast, but it's zero documented and developers tend to make major changes in it's design, so it's useless ,t oo08:44
B_LizzardOh well...08:45
B_LizzardAt least it exists08:45
kyakyes, that not so bad :)08:46
kristianpaulxiangfu: there?09:14
xiangfukristianpaul: yes09:15
kristianpaulnerase 16 512 0 009:15
wolfspraulqi never sleeps09:15
kristianpaulvs nerase 0 4096 0 009:15
kristianpaulxiangfu: the nerase in reflash script is suppposed to preserver something?09:15
kristianpaulof data*09:16
xiangfukristianpaul: no.09:16
xiangfuoh.. yes.09:17
xiangfuthe 0 to 16 is u-boot and kernel image.09:17
xiangfuit's nand block 0 ~ 16. which 0 ~ 7 is bootloader (u-boot),   8 ~ 16 is kernel09:18
xiangfu513 ~ END is data partition.09:19
xiangfu16 ~ 512 is rootfs partition.09:19
kristianpaulI'l add to the wiki09:19
xiangfukristianpaul: seems it's already in  wiki. let me find09:20
kristianpaulu-boot goes to page 0, Linux kernel to page 1024, rootfs to page 204809:21
kristianpaulbut you explain me it now in blocks, is nt?09:21
kristianpauli'm just flashing my ben again but dont want run nerase 0 4096 0 009:23
xiangfukristianpaul: checkout this http://en.qi-hardware.com/wiki/Hardware_basics#NAND09:25
xiangfu0 1024 2048 is nand page number.09:25
xiangfu0 8 16 is nand block number.   1 block = 128 pages.09:26
kristianpaulusefull !09:26
xiangfukristianpaul: you don't need to run "nerase 0 4096 0 0" but you need "nerase 16 512 0 0"09:26
xiangfukristianpaul: because if the .ubi file is small then your rootfs size, it will cause some kernel panic (cann't mount rootfs)09:27
xiangfukristianpaul: so if you reflash rootfs. always erase the rootfs partition.09:28
kristianpaulbtw anynews about run gcc on openwrt, is it posible now?09:29
wolfspraulkristianpaul: kyak worked on it09:31
kyakkristianpaul: works now.. however i haven't heard any feedback, but works for me :)09:32
wolfspraulkyak: sorry the packaging and making images department is slow, but we'll definitely get to it09:33
wolfspraulgcc doesn't get less useful over time... no worries09:33
kristianpaulok i can cross compile09:34
kristianpaulindeed wolfspraul09:34
kyakwolfspraul: np :) i'm not sure anyway it can fir into default image, because it occupies ~60 Mb with all libs09:34
wolfspraulkyak: we need to increase the rootfs partition to 512mb at least, I'm probably still leaning towards the entire 2gb but we will see09:34
wolfspraulall of this is sloooow :-)09:34
kyakit definitely should be split into several pacakges, because now everything is all together - static and shared libs09:35
kyakwolfspraul: might be a bad idea to occupy whole NAND with roots, because i use datafs as a "flash-safe" storage09:36
kyaksaving videos and music and things there...09:36
kyakbut it will be lost after reflash if rootfs occupies whole 2Gb09:36
wolfspraulsure no worries09:42
wolfspraulnobody listens to me anyway, it won't happen09:42
wolfspraulkyak: how about increasing the rootfs to 512 mb?09:42
wolfspraulwhen we change your data partition would be lost as well09:42
wolfspraulso the upgrade path is not good09:42
kyakthis seems reasonable to me09:43
kyakyeah, i understand.. but 512 Mb should fit for all our future needs09:43
kyak(oh, someone somewhere already said that)09:43
kristianpaulhmm why make menuconfig overwrite my .config file..09:50
xiangfukristianpaul: save old .config to .config.old09:50
kristianpaulah ok09:50
kristianpaulxiangfu: then run make?09:52
xiangfukristianpaul: I mean the "make menuconfig" will auto backup you .config to .config.old. (sorry )09:53
kristianpauli just wonder a .config file appear with other setup09:53
kristianpaulbut what i do to avoid that?09:53
kristianpauli want create an image using fullconfig09:53
kristianpauli run make V=99 then menuconfig is run09:54
kristianpaulis posible to avoid?09:54
xiangfukristianpaul: if the "make V=99" involve the "menuconfig" that mean there is no ".config" under openwrt rootfs folder.09:56
kristianpaulxiangfu: it is09:57
kristianpaulahh root fs??09:57
xiangfuwolfspraul: kyak. so we decide change rootfs to 512MB ?09:58
kristianpaulxiangfu: you mean in the root directory of my openwrt folder or in the files/ wich is the rootfs directory09:59
kristianpaulbut i have it..09:59
kristianpaulhm i forgot update feeds09:59
kristianpaullet see if that was the problem09:59
xiangfukristianpaul: root of opewrt folder, (like : /home/xiangfu/workspace/PanGu/openwrt-xburst )10:00
xiangfukristianpaul: sorry. it's root. not rootfs.10:00
kristianpaulis is there10:01
xiangfukristianpaul: try run this "./scripts/build"  this will backup you config file and stage_dir files.10:01
wolfspraulxiangfu: no not now, I think it's not full yet and we need to discuss with more people how to make the upgrade smooth10:04
xiangfuwolfspraul: ok.10:04
kristianpaulxiangfu: it works now after run update feeds, i think was my problem :)10:08
kristianpaulahh no10:08
kristianpaullets try your suguestion10:09
kristianpaulxiangfu: ./scripts/build is the recomended for creating openwrt images?10:10
xiangfukristianpaul: yes.10:11
kristianpaulah ok :)10:12
aisamorning everyone.  Last night a new user created a User page on the wiki:10:30
aisaI suspect this is spam.10:30
aisabefore I delete the page, I wanted a second opinion.10:30
kristianpaulyeah looks like10:32
wolfspraulaisa: no hesitations, I guessed I missed it :-)10:34
wolfspraulaisa: normally I lock the account (not the IP it came from), and delete the user page10:35
wolfspraulthanks for noticing and helping!10:36
aisawhere do I go to lock the account?10:38
aisaI found it.10:41
qi-bot[commit] Xiangfu Liu: remove useless functions in jz_mmc.c: http://qi-hw.com/p/openwrt-xburst/ee76fb410:58
wolfspraulaisa: great, thanks for the help! very good!11:02
rohhmmm.  anybody a idea where to find a appotech datasheet?11:47
rohfound a picture frame on sale for 40Eur with a 7" 800x840 screen and http://www.appotech.com/main/pro_detail.php?product_id=AX3003 as cpu11:48
qbjectHey all. I've got a BNN running OWrt as it was when I got it in August. Does switching to a smaller font still require compilation, per the FAQ?12:26
qbject(Hi, zear!)12:27
zearhey qbject12:27
zeari'm not an expert, but you shouldn't need to recompile just to change the font of the terminal. There was a command for that12:27
zearunfortunatelly i don't remember that command12:27
qbjectIf xdpirate shows up I'll ask him.12:29
qbjectzear: So, nethack-newt was a no-go on OWrt, even with additional libs.12:29
zearqbject, this could explain why most of the dingoo binaries don't run on jlime12:30
zeari guess it must be something with the core stuff, i don't know, glibc or something12:30
qbjectIs jlime compiled with uClibc?12:30
zearanyway, you could try jlime, it doesn't require flashing, it can boot from sd12:30
zearno idea, but probably yes12:30
qbjectOh! I hadn't thought of that. I'll get a space uSD some time soon and give it a try. Thanks for the suggestion.12:31
zearand how are the dingux libs you've transplanted? Does dingux stuff run on your nanonote now?12:32
qbjectNot everything.12:33
qbjectOpenTyrian and CDogs are both a bit slow, but run.12:33
qbjectThere's an interpreter for an old vector-rendered adventure game whose name I'll not utter that runs, if slowly.12:34
qbject(OpenTyrian is actually quite nice at lower detail settings.)12:35
zearthe thing is a lot of dingux apps convert the colordepth to 16bpp12:36
zearand nanonote then converts it to 32bpp12:36
qbjectNeedless overhead.12:36
zearso instead of 8bpp > 32bpp, it is 8bpp > 16bpp > 32bpp, so i guess that can also waste cpu12:36
qbjectI've heard that as well.12:36
qbjectWithout shoulder buttons I haven't found a good keymap for wolf/doom/quake, but they all go to some degree.12:37
zearqbject, you can try m/+, as they're close enough to the arrow keys to be easy to press in fast paced gameplay12:38
qbjectI'll give that a shit.12:41
zeari try to remind which ones i used12:41
qi-bot[commit] kyak: fbida: a few applications for viewing and editing images; ghostscript as http://qi-hw.com/p/openwrt-packages/277888f12:41
zeari think i used [qi] and ctrl12:41
qbjectOh, Meritous works well, too.12:46
qbjectFun and very strange.12:46
qbjectkyak: do you know if changing command line fonts on BNN OWrt still requires recompile?13:00
qi-bot[commit] kyak: a little help text from man fbgs http://qi-hw.com/p/openwrt-packages/ee8e1ef13:00
kyakqbject: it depends on how old your kernel is13:01
qbjectGot my bnn in August.13:01
kyaki don't know what's flashed inside.. you can just install setfont2 package and see yourself if it works13:01
qbjectThanks. http://amadeus.dist.ro/setfont/setfont.tar.bz2 ?13:02
kyaki suggest you flash this latest image http://downloads.qi-hardware.com/software/images/NanoNote/Ben/testing/2010-09-14/13:04
kyakit already includes setfont213:04
qbjectAhh. Thanks!13:05
mthzear: for 8bpp games, they should just an 8bpp surface from SDL and then SDL can convert it in one step to 16bpp on the Dingoo and to 32 bpp on the NN13:12
mth*just request13:12
mththat doesn't work though on booboo's kernel + rootfs13:13
mthfor 16bpp games, it might be useful if the NN kernel also had 16bpp support13:14
zearmth, i know, but most of the games were designed for booboo's kernel13:15
mthafaik the input pixel format and the output pixel format are independent, but I don't know the LCD controller very well13:15
qi-bot[commit] kyak: fbida: don't segfault on "h" http://qi-hw.com/p/openwrt-packages/2a7fda514:24
tuxbrain//msg NickServ tuxbrain akasha15:42
wpwrakssh root@tuxbrain.com  # let's see if he reuses passwords :)15:54
tuxbrainnot so easy15:54
wpwrakdarn :)15:55
tuxbrainwell whatever if you success to enter, please I have some manteinance task to do there heheeh15:55
wpwraktuxbrain: now there's an interesting "honeypot" concept :)15:56
bartbesI think  really need to set up a script to change some passwords on command15:56
bartbesjust in case15:56
qi-bot[commit] Werner Almesberger: Avoid leading zeroes in resistance and capacitance values (except 0R) http://qi-hw.com/p/eda-tools/274405716:31
wpwrakhmm, is anyone around here familiar with the FTDI chips in bit-bang mode, async or sync ? (not MPSSE)19:32
kristianpaulafaik i dont19:49
rohwpwrak not really. why?19:50
rohwpwrak there are some libs for speaking to that stuff19:50
wpwraki'm trying to send certain bit patterns. but the chip acts very very weird.19:50
wpwrakoh, sure. i'm using libftdi19:51
rohthe open one?19:51
rohhm.. i think thats the one openocd also used on moko, right?19:51
wpwrakyes, that's the same19:51
wpwrakwhat i'm trying to do is do the in-circuit programming of c8051f3xx chips19:52
rohthe examples worked for you?19:52
wpwrakthe examples are kinda simplistic :) one of the things i need is pulse timing control (for negative clock pulses)19:53
roheh... i dunno how well you can generate reproduceable timings below usb-latency witrh that chip19:54
wpwrakit has a self-clocked async and sync mode. either one should work for this19:54
wpwraki do get the right pulses. however, not always. even if i put a 1 s delay between packets, which should be more than enough to let usb get the thing out (and, besides, there's a fifo on the other end),19:55
wpwraki still see pulses that are too long. sometimes by a factor of 2, sometimes much more19:55
rohhm. could it be some usb-packet buffering foobar?19:57
wpwrakdata corruption could be a candidate. but then, i should see many completely missing pulses. also, i send on two channels. if there's data corruption, i should see occasionally one channel differ from the other.19:57
wpwrakso it does seem to be the underlying timing that's off19:57
wpwrakthe chip has a 256 bytes transmit fifo. i send two bytes every second. worst case, some pulses could get packet back to back. but i can't quite imagine a fifo overrun.19:58
wpwrakbesides, also then, i should see this in the timing19:58
wpwraki mean the time between pulses19:58
wpwrakit could be that usb i/o in general upsets the chip, but that would be pretty weird. oh, and i checked the power supply as well. it's perfect.19:59
wpwrakthe chip (ft232rl) has an internal rc oscillator. so there's no crystal or pll to worry about. according to documentation, the internal oscillator is free-running, i.e., not disciplined by the usb clock. (unlike, say, the silabs chips)20:01
Action: kristianpaul reads " This slide contents is only available to the listeners of courses"21:52
kristianpaulheh seems lot of people is making money now teaching how to decode GPS signals :p21:53
wpwrakkristianpaul: you'll be rich ! :)21:56
kristianpaulwpwrak: not yet, first it need make it work here22:08
wpwrakkristianpaul: small but important detail :)22:11
kristianpaulxiangfu: do you have at hand the rootfs image used in the 65 SIE boards?22:13
kristianpauli need reflash mine, soemthing is wrong with kernel modules..22:13
wolfspra1lkristianpaul: xiangfu never held a SIE in his hands, just fyi22:14
kristianpauland adam pointend in the wiki that for my board the specific problem was solved after reflash rootfs22:14
kristianpaulwolfspra1l: oh ok22:14
kristianpauli wonder if the rootfs image came from Carlos then?22:15
kristianpaulhe should i think btw :)22:15
wolfspra1lwell we have no boards22:15
wolfspra1land it's all a matter of priorities22:15
kristianpaulsure sure (MM One RC2)22:15
wolfspra1lI hope there is nothing wrong with your board, but for now I wouldn't think so.22:16
wolfspra1lour testing and quality ensurance is very weak for SIE22:16
wolfspra1lI wouldn't know why 'reflashing twice' should fix anybody's bug, but it did for Adam two or three times among the 65 boards22:16
kristianpaulwell i was about simualte some signaling when discover the sound modile in kernel was gone..22:16
wolfspra1lyou mean the file in the rootfs is just gone?22:17
kristianpauli just eant reflash it22:17
kristianpaulxiangfu: compiling backfire with minimal image i got this http://paste.debian.net/97178/22:18
xiangfukristianpaul: the Ben NanoNote rootfs should works in SIE.22:19
kristianpaulso no bother for get the minimal one?22:20
xiangfukristianpaul: yes.22:23
wolfspra1lxiangfu: you mean SIE uImage + NanoNote rootfs?22:24
wolfspra1lor also NanoNote uImage?22:24
xiangfuwolfspra1l: yes22:24
xiangfuSIE uImage + NanoNote rootfs22:24
xiangfuwill works fine.22:24
kristianpaulactually SIE overwrite some of the nanonote rootfs but is okay22:24
kristianpaulwait a min22:24
kristianpauldamm i send the wrong uImage..22:25
xiangfukristianpaul: http://projects.qi-hardware.com/index.php/p/nn-usb-fpga/source/tree/master/binaries/sie_rootfs_files22:25
kristianpaulsure i have those22:26
xiangfuhere is I try to clean up the SIE rootfs. it's only add some fpga command.22:26
kristianpaulwhat happen if i flahsed nannote uImage...22:26
kristianpaullets wait this rootfs finish and i'll tell22:26
xiangfukristianpaul: some kernel panic :)22:27
kristianpauli think i'll write an email to carlos asking for rescue files ;)22:27
wolfspra1lkristianpaul: NanoNote uImage definitely cannot boot on SIE22:29
wolfspra1llike I said someone needs to cleanup the patches and create a proper OpenWrt target for it22:30
kristianpauluImage is just kernel isnt?22:30
wolfspra1lyes, without modules22:30
kristianpaulwolfspra1l: indeed this is all a mess (sight)22:30
kristianpauli wonder how carlos students can survive this ;) (sure they will learn a lot)22:31
kristianpaulit booted !22:32
kristianpaulno panic22:32
kristianpaulso far..22:32
wolfspra1lkristianpaul: you mean with Ben NanoNote uImage?22:39
wolfspra1lwell some things will not work22:40
wpwrakwow, digi-key are better than i knew. 16:50 (north dakota time): order submitted. 18:32: order ready for shipping. 19:13: data sent to fedex. 15:53: shipment picked up. no i know how they can be so fast. darn time travelers !22:46
wolfspra1lone item may come from fedex, which may use a different timezone that is lost in the transfers somewhere?22:47
wpwraknew, the timezones normally are right. the last two times are from fedex. i guess the pickup truck needs a clock adjustment :)22:49
brantaris this the right place to ask about rfid?23:17
wolfspra1lbrantar: nice to see you, what takes you here?23:17
wolfspra1lhe. very remotely. we discussed a project with rfidguardian once but it never went beyond the telephone call stage :-)23:17
wolfspra1lbut your time is not wasted at least asking your question here, so go ahead23:18
brantarwolfspra1l: ahh, im looking to connect a Parallax serial rfid reader to my pc for logging in/entering passwords/etc.23:19
brantarwolfspra1l: my first hurdle is the physical connection though... see, i am no electronics guru23:19
wolfspra1lneither am I, but miraculously I ended up in this hardware project...23:20
brantarwolfspra1l: Haha!  i guess were in the same boat, eh?23:21
wolfspra1lI'm afraid I have to pass. Are you talking about normal serial (uart)? Maybe just a level-shifter or usb2serial cable will help you? are you talking about wiring cables or just plugging connectors?23:22
wolfspra1lif we do an rfid project (which I don't see right now), we would try to make it a really open project, along the lines one of our projects called ben-wpan (which is using 802.15.4)23:23
brantarwolfspra1l: well with my limited knowledge i do know i can buy a usb2serial convertor... but i would like to avoid costs if at all possible23:23
brantarwolfspra1l: who is the "we" you speak of?23:24
wolfspra1lanybody who likes copyleft hardware23:24
wolfspra1las secondary citizens, we also allow people who merely like open hardware23:24
wolfspra1l(that's not serious, just in case... :-))23:24
brantarwolfspra1l: so you're talking about making projects a community thing on here? i'm just looking for a little electronics help23:24
wolfspra1lyes not sure someone can help you, maybe tuxbrain? (he's probably sleeping now)23:25
lisandropmbrantar: let me read the backlog23:25
wolfspra1lcurious now, what happened to rfidguardian, let me check... :-)23:26
lisandropmbrantar: wich kind of interface does the parallax has?23:26
brantarlisandropm: I just wanted to know if i could connect http://bit.ly/bchBk8   to   http://bit.ly/a5hwoB23:26
lisandropmlet me see23:26
lisandropmbrantar: I can't see if the rfid is rs232 or not, but I don't really think so23:27
brantarlisandropm: I have seen people on the forums on there talking about connecting it (through arduino's and such) to an RS232 connector23:28
brantarlisandropm: the problem: my computer has no RS232 ports23:28
lisandropmyes, but I don't know if it needs a traslator or not23:28
brantarlisandropm: where would i find that?  btw, the representative at Parallax said this would work to scan to my pc... but im not sure if he just wanted to push another product or was telling the truth.23:30
lisandropmbrantar: do you have a datasheet for ot?23:30
rohbrantar looks fine for me23:30
lisandropmfor me too23:30
brantarroh: lisandropm: fine as in...?23:30
lisandropmfine as in it should work23:30
rohgive the reader power, tie the enable line to ground and the tx to the rx of the usb-thingie, also ground23:31
lisandropmsee, the question here is wich voltages does the interface uses23:31
rohsee if it works23:31
rohthe rfid reader uses cmos/ttl compatible 5V logic levels. as the parallax usb thingie is also compatible to23:31
brantarroh: enable and ground to the grounding of a usbcable or something?23:31
brantarroh: would it need to be plugged in?23:32
lisandropmroh: nothing to add then ;)23:32
rohto the gnd of the power supply and to the gnd of the usb thing23:32
rohthe VSS on the usb thingie means ground23:33
brantarroh: again, im slow lol.  so the reader would have the GND going to? and the enable going to?23:33
rohignore res and the tx23:33
rohyou only need rx and the vss.23:33
rohconnect rx to the pin3 (sout) of the reader23:34
rohconnect vss to pin 4 (gnd) of the reader23:34
rohconnect pin2 (_enable) to pin4 (gnd) of the reader23:34
rohconnect 5v (from some power supply) to pin 1 (vcc) of the reader and gnd of the supply to the pin4 (gnd) of the reader23:36
rohyou could also use another usb port as supply, but then you need to be careful not to generate any shorts23:36
brantarroh: such as a plugged in usb  cable?23:36
rohtrue. the usb-thing from parallax has only 4 pins and no vcc line as it seems.23:37
brantarroh: so all it does is transmit data, but not grounding or power?23:38
rohi'd just use my lab psu or cut a second usb cable. isolate the 2 datalines and use red and black for power23:38
rohbrantar the usb thingie recieves data from the reader23:38
brantarroh: and that is all it does?23:38
rohits also quite easy to hack rfid stuff (125khz transponder tags)23:39
roheasy to clone/emulate23:39
brantarroh:  ok, so i need to draw a diagram to see if i got it right.23:39
rohnot good for security stuff ;)23:39
brantarroh: how so?23:39
rohits uncrypted tags.23:39
roheverybody could read them out and clone a card or emulate one.23:40
brantarroh: but someone would have to have physical access to my tag to do any damage right?23:40
rohor be near you at some point (stand next to you in an elevator or so)23:40
brantarroh: im not withholding state secrets with this tag haha.  im just looking to keep a blank tag on me that can log me in to  my pc and such23:40
rohsometimes even some distance can be circumvented23:41
roh(using better antennas)23:41
brantarroh: but for the average high school student i shouldnt worry too much?  :D23:41
rohi wouldnt use it instead of my pw.23:41
rohand it depends... on what kind of school you are on. on some you find quite whitty people;)23:42
brantarroh: ok, ill keep that in mind.  how far of a distance do u think would be readable for an implanted tag?23:42
rohdunno.. i only know of card-stuff23:42
brantarroh: how about for card then?23:42
rohit depends on the power used to read.23:43
brantarroh: ok.23:43
rohone thing it reading it with another reader, another is eavesdropping on your reader23:43
rohthe first one means generating the power-field yourself, the second is completely passive (no tx), and only listening23:44
brantarroh: ok, well ill probably just ue it for automating stuff then, maybe starting my common programs or something23:44
brantarroh: thanks for the immense help, im going to draw a diagram to make sure i got all you said correct.23:45
wpwrakwolfspra1l: btw, are there any schematics/layout of the breakout board ?23:50
wolfspra1lhmm. if at all then in ben-blinkenlights?23:52
wpwraknope, nothing there23:54
wpwraki mean the board that connects to a cable etc. you said that you're already producing it ?23:55
brantarroh: SORRY, im slow as molasses, almost done with the schematic23:57
wolfspraulhe, not really a controlled 'production', just a street job in Shenzhen I need to follow up on it... don't know where it stands.23:57
wolfspraulI think I will try to produce some mmone-jtag-cables in a more controlled way, that could be a nice test for boom too, for example23:57
wpwrak(street job) oh :) well, let's see what comes out of this ;-)23:58
wpwrak(mm1-jtab) yeah, that should be nice and simple23:58
--- Tue Oct 19 201000:00

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