#qi-hardware IRC log for Tuesday, 2013-10-29

sanderrThat's cool! But how open is open?12:21
whitequarkI was more thinking along the lines of phoneblocks12:22
whitequarkwhich was the same thing but the consensus here was that it's dommed12:22
whitequarkI'm still not convinced of technical feasibility but I wonder what is Motorola thinking12:23
sanderrI'm reading the phonebloks website.12:25
sanderrIt's a reaction to the disposable world, and it emphasizes more on the sharing and app economy.12:25
sanderrBut I like the idea.12:26
sanderrwhitequark: I think Motorola wants to move into a niche market.12:26
whitequarksanderr: I'm not sure Motorola wants to do anything. it's just their R&D division.12:32
whitequarkAnd I wonder how does that R&D division wants to solve stuff like mechanical wear, exceeding cost, slow interfaces, ...12:32
sanderrCheap (and probably not fully standard) copy parts.12:46
wpwrakso they now have invisible pink unicorns at the ivory tower ...12:51
wpwrakat least this speaks well of the financial situation of googlorolas blue sky research division. but they could try to be a bit more original. that ideas's been floated before, with very similar cad work.12:52
wpwrakor one could say "cgi" or "sfx" instead of "cad". makes it sound a bit less serious :)12:54
sanderrI like the way you put it. "googlorolas blue sky research division" :) But it's not very original. It's like Lego building blocks, but with a CAN (automotive) bus.12:55
wpwrak"After the trip, we asked ourselves. [...]" hmm yes, that seems to be a not so uncommon meme, too :)12:55
wpwraksuch concepts have been proposed several times already. what they have in common is that they don't really make sense.13:07
wpwrake.g., if your goal is upgradeability, then you'll find that, in, say, two years, you'll not only want a faster CPU (really ?), but also a new modem that supports a new band that's been added in your area13:08
wpwrakand when you touch the modem, you're in the middle of RF compatibility issues. so that's the least modular part of all.13:10
sanderrSo all in all, the concept is perfect on paper, but useless in the physical world?13:26
larscI wouldn't call in perfect on paper13:28
wpwraki'd say so, yes. better flexibilize the manufacturing process and move from mass production to build-to-order. then you can have modularity at a higher level.13:28
wpwrakheh ;-)13:28
wpwraklet's say that it looks pretty on paper :)13:46
whitequarkthat's for sure13:48
DocScrutinizer05any network cracks around?14:07
DocScrutinizer05I'd need some help on debugging a weird behaviour on 3 of my WLAN devices14:07
DocScrutinizer05DNS resolving doesn't work enamore after several hours of device "idling"14:08
DocScrutinizer05my network topology is a bit unusual: I got two NAT/routers between WLAN and the internet14:09
DocScrutinizer052nd level router (the one giving DHCP leases) is set up to hand out sttaic public DNS server IPs14:09
DocScrutinizer05/tmp/resolv.conf.wlan (dnsmasq specific way to handle DNS) has the correct public DNS server IPs on a device that refuses to ping ct.de14:11
DocScrutinizer05("unkown addr")14:11
DocScrutinizer05disconecting/reconnecting WLAN fixes the issue for a day or so14:11
DocScrutinizer05the PC - connected via cat5 - on same subnet never experiencing any such issue, I can't rule out it's due to PC never "idles"14:14
sanderrDocScrutinizer05: Does your wlan go into sleep mode?14:14
DocScrutinizer05err sleep mode? it does PSM yes14:15
DocScrutinizer05I never shut it down though14:15
DocScrutinizer05to me it actually seems like NAT sessions closing eventually14:16
DocScrutinizer05once device that is running IRC never goes offline for IRC, and never goes offline for ssh session. It however every now and then goes offline for SIP :-S .oO(???)14:17
DocScrutinizer05and moreoften than not I find an occasional web browser I open telling me "server not existing"14:18
sanderrSeems that it closes 'stale' network connections after a given time period.14:19
DocScrutinizer05when I do a ping ct.de then, it also fails. Then I unregister/reconnect WLAN and everything (browser, SIP, ping) works again14:19
DocScrutinizer05yeah, but shouldn't the local dhcpcd or whatever open a new connection whenever needed?14:19
sanderrAnd if you keep pinging it in the background every 60 seconds? (as in, keep the connection open?)14:19
sanderrDocScrutinizer05: I guess it should.14:20
DocScrutinizer05route looked good14:20
DocScrutinizer05/tmp/resolv.conf.wlan looked good14:21
DocScrutinizer05ping works14:21
sanderrbut your network map is: inet --> router1 -(dhcp)-> router2 -(static)-> device 14:21
DocScrutinizer05ping ct.de fails14:21
sanderrCan you still ping router2?14:21
DocScrutinizer05I even can ping the intarwebs14:21
sanderrYou should first find out which connection is dropped, the dhcp-one or the static one.14:21
DocScrutinizer05it's strictly DNS that fails14:21
sanderrWhat dns servers are given out?14:22
DocScrutinizer05plus of course everything that depends on DNS14:22
sanderryou can set up and on the client systems as dns servers.14:22
DocScrutinizer05the statically configured ones14:22
sanderrIt's a workaround, but might work.14:23
DocScrutinizer05I have statically configured DNS IP addr on router level214:23
DocScrutinizer05I don't see the difference when I configure local DNS IP 14:23
DocScrutinizer05the ones listed in /tmp/resolv.conf.wlan were exactly those I configured in rouer-l214:24
sanderrAnd you could still ping those?14:24
DocScrutinizer05so it seems DHCP works, from device POV14:24
sanderrSo there was a cutoff on port 53 (dns).14:24
DocScrutinizer05THAT is a good qustion :-)14:24
DocScrutinizer05I only pinged 8.78.814:25
sanderrthat's the one I always ping too. :)14:25
DocScrutinizer05but not my DNS I'm using14:25
sanderrtry that next time, and try to find out how far you can get locally14:25
DocScrutinizer05I wonder if friggin fritzbox (L1-router) is intercepting port5314:26
sanderrewwww... a fritzbox14:26
DocScrutinizer05not my decicion, not my domain14:26
DocScrutinizer05I can configure it, but that's it14:27
sanderrare there by any chance sip-clients behind it?14:27
DocScrutinizer05I can't even install tellnet/ssh on it14:27
DocScrutinizer05sure, mine14:27
DocScrutinizer053 or 4 ;-)14:27
DocScrutinizer05(SIP clients)14:27
sanderrbecause what I know, fritzboxes prioritize their own (on the hardware) sip to be "more equal" than others (connecting through it)14:28
DocScrutinizer05and SIP pretty much acting up since that friggin fritzbox came into my path14:28
sanderrI know, they're a pain14:28
DocScrutinizer053210, no SIP but seems some ALG14:28
DocScrutinizer05I can't pinhole 5060 -> "internally used, sorry"14:29
DocScrutinizer05but that's kinda plausible. What however is NOT is the DNS fsckup14:30
DocScrutinizer05it's goddamn annoying14:30
sanderrsorry that I can't help any further, but you need to investigate a bit more14:30
DocScrutinizer05and strange14:30
DocScrutinizer05thanks for the (basically obvious, but i missed it) hint to ping my actual DSN servers14:31
sanderrconnect your level2 router with utp to the fritzbox?14:31
sanderrwell, fingers crossed!14:32
DocScrutinizer05before there was no fritzbox in this picture, and everything fine14:32
DocScrutinizer05err what?14:34
DocScrutinizer05I miss the context there14:34
DocScrutinizer05seems like answers to an unknown OP14:34
DocScrutinizer05ohnoes, phonebloks again14:36
sanderrDocScrutinizer05: That's on Phonebloks and Motorola/Google going at it again.14:36
sanderrWe discussed that before the Fritzbox debacle. :s14:36
DocScrutinizer05seems an idea needs a certain minimum of *idiocy* to really kick of a hype14:37
DocScrutinizer05off, even14:37
DocScrutinizer05well, when they increase the size of phoens err phablets further at same rate, we soon will be able to use PCIe cards14:40
DocScrutinizer05THEN the whole modularity idea starts to become feasible, thoufg still nonsensical14:41
DocScrutinizer05after all, look what PCIe card you got in your current PC (if you're still using a PC and not a laptop already)14:42
DocScrutinizer05and particularly which cards are non-existent: decent coproc cards. RAM cards. Even decent GFX cards (need PCIe-x)14:43
DocScrutinizer05but hey, maybe they can use a backplane (thry call it 'base'?) full of USB3 hubs ;-P14:45
whitequarkall your base are belong to us14:46
DocScrutinizer05speedevil had a nice characterization for the phonebloks idea14:46
cdewhitequark: this joke expired 5 years ago14:46
DocScrutinizer05[2013-10-25 16:03:44] <SpeedEvil> My response in another place - http://permalink.gmane.org/gmane.comp.handhelds.openmoko.community/63678 This is basically at the level of a 5-year-old trying to design a car.14:49
DocScrutinizer05<SpeedEvil> 'Ok - it needs wheels and doors and a ball-pit'. Without the knowledge  of what the transmission or suspension is.'14:49
DocScrutinizer05>>Can it be done - sure!14:53
DocScrutinizer05Will it be twice the price, twice the weight, a quarter of the reliability - very likely.<<14:53
DocScrutinizer05I disagree. Very likely it won't fly at twice the size and price. I estimate at least 8 times the volume and 20 times the price14:54
DocScrutinizer05just like those jolla-TheOtherHalf freaks suggesting to implement dual-SIM by simply placing a complete 2nd phone into TOH14:57
DocScrutinizer05even if it could be done, you run into a hell of a RF-design nightmare, since two aerials won't work nice in such close neighbourhood to each other14:58
wpwrakbtw, for those not reading the list, here's a bit of y-box fun: schematics: http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybox-20131025.pdf15:34
wpwrakand layout: http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybox-20131025.png15:34
wpwrak(y-box is the critter that combines a powerless usb host with usb power from some pc/wallwart/etc. and lets some usb device (keyboard or similar) plug into all this.15:35
DocScrutinizer05those eagle schematics always are terribly ugly15:38
DocScrutinizer05or is this KiCAD even?15:38
wpwrakof course it's kicad15:39
wpwrakand what's ugly about it ?15:39
DocScrutinizer05dunno, the ground symbols, maybe the font, the colors, and particularly any wire ending in nirvana, with only a (electrically) irrelevant writing indicating what might be the deal with it15:41
DocScrutinizer05a wire from a grounf symnol to a box labeled [POWERD]15:42
wpwrakthe latter is rather common if drawing the lines would be messy15:42
DocScrutinizer05not in any professional schematics I ever seen15:43
wpwrakwell, POWERED is one of the less pleasant kicadisms, i agree. don't like it either, but it keeps the (otherwise useful) ERC complaints away.15:43
DocScrutinizer05also not in DIN schamics symbol definitions afaik15:43
wpwrakthe POWERED ? i'd be surprised if it existed in any standard15:45
DocScrutinizer05no, wires ending in nirvana without *any* termination15:45
DocScrutinizer05a decent CAD wouldn't allow such stuff fist instance15:46
wpwraknaw, a real CAD checks what things you are trying to connect15:46
DocScrutinizer05you either use a bus symbol or an offsheet symbol to terminate the wire15:46
DocScrutinizer05offsheet symbols look like / similar to your [POWERED], and have to come in pairs15:47
DocScrutinizer05as they are virtual invisible wires15:48
wpwrakanyway, that's not the issue here. one thing that gives me a bit of a doubt is whether the micro-USB connectors should place the plug "up" or whether it doesn't really matter.15:48
wpwrakright now, i'm using non-reversed receptacles, which are intended for bottom-side mounting. since i use them on the top, the plug thus faces "downward"15:49
DocScrutinizer05I dunno if anybody actually cares15:49
wpwrakthe plug's overmold has a top and a bottom side. the top side may get a USB logo or at least "A" and "B". the bottom side may be blank or have a vendor logo.15:50
wpwrakyeah, that's my question :)15:50
DocScrutinizer05well, *I* know that15:50
DocScrutinizer05I dount anybody else does15:50
DocScrutinizer05and even I don't 'use' that knowledge15:50
DocScrutinizer05usually users do the try-and-error method on mating plugs15:51
wpwrakyeah, the procedure is pretty much: take plug, try to fumble it into the receptacle, if it fits, done, if it doesn't, reverse and repeat, if it still doesn't fit, have a closer look15:51
wpwrakindeed :)15:51
wpwrakit gets a bit nastier with AB receptacles because they accept the plug in either orientation and it's only the plastic and not the metal preventing it from entering the wrong way15:52
DocScrutinizer05I actually look at the *plug* of my usb charger before plugging it into the N900. But not at the symbols in plastic, but at the actual metal sleeve15:53
wpwrakthat's not a problem in this circuit but the password safe itself has an AB receptacle15:53
wpwrak(to make things more interesting, the china-made micro-B cable i tried to attach has plastic from the overmold spilled under one of the lock "tongues", so even if inserted correctly, it was almost impossible to push it in, let alone get it out again)15:54
DocScrutinizer05need to hurry15:54
wpwrak(short-term solution: cut off that "tongue". now it moves properly. long-term solution: accept the lesson that they still don't know how to make decent connectors in china)15:56
Action: whitequark switches to the channel and reads "cut off that tongue"16:04
whitequarkshould I read backlog?..16:04
wpwrakit's a piece of metal in a connector :)16:06
wpwrakno preventive whistleblower treatment :)16:06
--- Wed Oct 30 201300:00

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