DocScrutinizer05 changed the topic of #qi-hardware to: Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs and http://irclog.whitequark.org/qi-hardware
<wpwrak> heh, there's also a clone from .ve. a bit easier than yours: http://jotase.github.io/FlappyFree/#2
<wpwrak> note that they have some currency restrictions there the people really resent ;-)
<wpwrak> here's a collection of clones: http://www.palmentieri.it/somepx/flappyjam/
<dos1> wpwrak: haha, submited :D
arielenter has quit [Quit: Leaving.]
wolfspra1l has joined #qi-hardware
<whitequark> DocScrutinizer05: I see, thanks
<DocScrutinizer05> hm?
<DocScrutinizer05> soory, stack cleared
<whitequark> re grounding and wristband
<DocScrutinizer05> aah, np
<whitequark> wpwrak: (salt) I'd search pubmed to substantiate my claim but I'm too lazy right now :p
<DocScrutinizer05> a very simple tip: use a wooden desk
<whitequark> DocScrutinizer05: I already do
<DocScrutinizer05> good
<whitequark> well, it's some kind of wooden laminate
<whitequark> but that should work just as well, wouldn't it?
wolfspraul has quit [Ping timeout: 248 seconds]
<DocScrutinizer05> second tip: try to always first touch the desk's metal feet before you touch anything sitting on the desk
* whitequark nods
<DocScrutinizer05> even better would be a metal desk of course
<DocScrutinizer05> 3rd tip: always touch electronics at the ground plane or some metal frame first
<DocScrutinizer05> it's usually not the electronic device that's electrically charged, it's usually you
<DocScrutinizer05> ESD concerns in lab are massively exaggerated nowadays, since 99.9% of components already have quite decent ESD protection integrated
<whitequark> DocScrutinizer05: I'm quite bothered when I stand up and hear a loud bang from the discharge
<whitequark> like, it's the kind of bang you hear when you plug something with 10000uF cap into mains
<DocScrutinizer05> that's bad! you need to do something, change the situation, eliminate the van der Graaf
<whitequark> hehe
<DocScrutinizer05> laminate may have a very high part of resin resp epoxy or plastic in general
<DocScrutinizer05> your chair shouldn't be plastic either
<whitequark> wpwrak: (exif) imgur strips exif data. likely because it's the default image host for reddit, and unknowingly posted exif is a huge source of personal data leaks
<DocScrutinizer05> a desk lamp with PE-grounded metal case that you touch to switch it on, each time you sit down at your desk, may help wonders
<DocScrutinizer05> but when you hear static noise when you stand up, then you got the van der Graaf locally at your desk, and then you *really* need to fix that
<DocScrutinizer05> I's blame your chair
<whitequark> I'm pretty sure it is the chair
<DocScrutinizer05> and your desk surface together with the clothes on your forearms
<whitequark> nah, no clothes on forearms--using a tshirt
<DocScrutinizer05> ok, very good
<DocScrutinizer05> use a conductive "blanket"
<DocScrutinizer05> on your desk
<DocScrutinizer05> way better than wrist band
jekhor_ has quit [Ping timeout: 250 seconds]
gbraad has joined #qi-hardware
gbraad has joined #qi-hardware
gbraad_ has quit [Ping timeout: 272 seconds]
whitequark has quit [Ping timeout: 264 seconds]
whitequark has joined #qi-hardware
dos1 has quit [Read error: Operation timed out]
<DocScrutinizer05> whitequark: yep
<DocScrutinizer05> get one that's large enough to not annoy you
<DocScrutinizer05> I.E. one that's so large that you don't even think of placing something _not_ on the mat
<DocScrutinizer05> of course a metal plate works as well
xiangfu has joined #qi-hardware
unclouded has quit [Ping timeout: 245 seconds]
luke-jr_ has joined #qi-hardware
Luke-Jr has quit [Read error: Connection reset by peer]
qwebirc72587 has joined #qi-hardware
bzb has joined #qi-hardware
luke-jr_ has quit [Remote host closed the connection]
luke-jr_ has joined #qi-hardware
qwebirc72587 has quit [Quit: Page closed]
bzb has quit [Quit: I hate to quit but my bladder's full. :-(]
luke-jr_ is now known as Luke-Jr
arielenter has joined #qi-hardware
unclouded has joined #qi-hardware
arielenter has quit [Remote host closed the connection]
arielenter has joined #qi-hardware
arielenter has quit [Read error: Operation timed out]
arielenter has joined #qi-hardware
arielenter has quit [Quit: Leaving.]
wolfspra1l has quit [Ping timeout: 246 seconds]
jekhor_ has joined #qi-hardware
wolfspraul has joined #qi-hardware
wolfspraul has quit [Quit: leaving]
wolfspraul has joined #qi-hardware
rz2k has joined #qi-hardware
lekernel has joined #qi-hardware
xiangfu has quit [Quit: leaving]
<apelete> Hi larsc
wej_ has joined #qi-hardware
<apelete> larsc: did you receive the email of my documentation patch ?
<apelete> larsc: there seems to be encoding issues on the patch, so I was wondering what I did wrong :-(
wej has quit [Ping timeout: 252 seconds]
<larsc> apelete: I got the mail. What kind of encoding issue?
<apelete> larsc: some characters does not show up: http://marc.info/?l=linux-doc&m=139199386820719&w=2
<apelete> lkml doesn't even show the email at all: https://lkml.org/lkml/2014/2/9/285
<apelete> larsc: but it's ok if you got the mail, that means it's a client issue
<larsc> it looks good in my mail client
<apelete> ok, thanks for the feedback
<apelete> larsc: with the documentation done, I can now get back at debugging the dma in the mmc driver
<apelete> expecting a painful experience, does not even know where to start :-(
<larsc> should be that bad
FDCX has quit [Ping timeout: 252 seconds]
<CYB3R> Is it possible to boot something on Xburst from USB (without flashing to NAND)
<CYB3R> ?
FDCX has joined #qi-hardware
<larsc> yes
<CYB3R> larsc: what commands should I pass to jzboot?
<larsc> that I don't know
<larsc> has been a long time that I used it
<larsc> for anything else but flashing the nand
<CYB3R> Is it possible to use UART or GPIO without RAM?
<larsc> yes
<larsc> if your program fits in the OCM
<larsc> aka IRAM
<CYB3R> I will be soldering a board in a day or two, so I want to test if everything works
<CYB3R> Maybe I'll be doing tests before soldering NAND and SDRAM
<nicksydney> wpwrak: ping ~
<wpwrak> qi-hw emergency response team. please state the nature of your emergency.
<apelete> XD
<larsc> not very responsive though that response team
<wpwrak> well, the caller hung up ...
rz2k has quit []
<CYB3R> just got awesome TCA8424RHAR from TI
arielenter has joined #qi-hardware
dos1 has joined #qi-hardware
dos1 has quit [Changing host]
dos1 has joined #qi-hardware
<larsc> yeay, systemd *celebrate*
<whitequark> you mean *mourn*
lekernel has quit [Ping timeout: 260 seconds]
<larsc> I celebrate
<wpwrak> whitequark: he means this kind of celebration: http://www.clipartpal.com/clipart_pd/holiday/halloween/halloweendance_10145.html
<larsc> switching to upstart would be switching from a bad design to a bad design
<larsc> switching to systemd is switching from a bad design to a good design
<whitequark> you can't seriously call systemd a good design
<whitequark> I could kinda accept "we don't currently have resources for yet another better init replacement"
<whitequark> which we, unfortunately, likely don't
<larsc> I can ;)
<whitequark> then you're a troll
<larsc> no
<wpwrak> that's a good crowd. once they get to the vodka, the celebration will be fun
<wpwrak> (troll) or maybe he's poettering ? :)
<roh> larsc: debian decided?
<roh> nice
<roh> whitequark: good designs can be reimplemented properly and become better. bad designs stay shit regardless of the number of implementations.
<whitequark> roh: exactly, systemd can't be fixed.
<kyak> whitequark: why is systemd bad design?
lekernel has joined #qi-hardware
<whitequark> kyak: it's a monolithic thing which tries to do everything at once. from running services to managing networks to udev to tracking console sessions to displaying qr codes
<kyak> well.. is it bad?
rz2k has joined #qi-hardware
<whitequark> you'd think so
<whitequark> larsc: okay, after looking at 20 and 22 I'll maybe reconsider it.
<roh> whitequark: upstart can't be fixed
<roh> and sysv is through.
<roh> systemd can be fixed. atleast its a sane concept cleaning up all the sins of the stone-age a bit better
<dos1> the biggest pain of systemd for me is that it's Linux only
<roh> its not perfect for sure. but its a change, which i think can help us out more than upstart and sysv together.
<dos1> but I guess that could be fixed
<roh> also note.. for embedded you usually use something different.. like openwrt
<whitequark> dos1: actually that one probably could not. lennart said this explicitly, even
<larsc> it's very hard to fix
<whitequark> "too much linux-specific interfaces"
<roh> openwrt uses its own replacement: procd and its companions netifd and some more.
<larsc> you'd have to port stuff like cgroups to $TARGET_OS
<roh> whitequark: i could not care less. yes its linux specific so what.
<dos1> I don't mean systemd implementation, but the design
<whitequark> roh: tat was to dos1
<roh> solaris has another, own implementation of all that, osx too.. so why care?
<whitequark> *that
<roh> whitequark: sorry
<whitequark> dos1: implementing it is a huge deal though... since it's so complex
<roh> yes.. dos1. doesnt matter.
<dos1> I believe that good design should be at least somehow portable
<roh> dos1: no.
<dos1> if POSIX lacks some features, extend it and provide well-defined standard, so anyone else can implement it if they want
<roh> good designs solve proplems. defined problems. when you define all problems into one, your design will be bad and never be implemented properly.
<roh> and posix is part of the problem, not the solution.
<whitequark> dos1: you forget though that POSIX is defined after the fact
<roh> see gnu hurd. posix to the death doesnt dolve anything.
<whitequark> look at POSIX AIO. no one implements or uses it.
<roh> forget classic unix. thats whats holding us back. yeah there are some good parts on that concepts. those we keep. but the legacy mindfuck can go and stay away
<whitequark> I'm not even sure how POSIX is useful for anything except extorting money for certifications
<roh> in the future one will have linux with systemd and ipc via kdbus.
<whitequark> because there's as much subtly incompatible implementations as there are kernels
<dos1> from my point of view, if there's some well documented standard (call it even "the Linux standard") I can conform to when writing my own kernel and that will be enough to make systemd working on my kernel, then it's "portable enough"
<whitequark> as many*
<whitequark> dos1: you'd be basically rewriting implementation details of Linux. that's the problem.
<dos1> I don't care if this standard is in fact POSIX, Win32 or anything else
<whitequark> what is the point of implementing your own kernel just to badly emulate Linux with it?
<dos1> whitequark: yes. that's why I want well defined and documented standard, not "Linux emulation"
<whitequark> it's a chicken-and-egg problem. you need competing implementations for a sensible standard to emerge
<dos1> you might be right
<whitequark> and I honestly don't think that any kernel is going to compete with Linux in... a decade or something? at least. that's how long it takes to set things up
<roh> dos1: there is nox _fixed_ standard
<roh> you are always dependant on behaviour and features of the state of versions you code against.
<dos1> but I fear that it will stop on "custom implementation" stage, exactly for that (I honestly don't think that any kernel is going to compete with Linux in... a decade or something?) reason
<roh> so YES, compiling something against a glibc2.something makes one depending on that or later compat.
<roh> but hey... you COULD just do opensource or recompile your crud from time to time. and be fine.
<whitequark> you don't get standards by thinking "oh we need to make a standard for it"
<roh> compatibility is something one can have cheap and easy.. on the source level.
<whitequark> that's how you get "standards" like that shit microsoft creates. it solves only the use case of the person writing it
<dos1> source level is enough for me
<whitequark> you need actual competing implementations people would use
<roh> the kernel will never be complete
<whitequark> and collaboration
<whitequark> a lot of it
<whitequark> see: browsers
<roh> linux is an iterative, evolving process of an os. nothing which will ever be done, complete or finished, i think
<whitequark> is software ever complete? :)
<dos1> whitequark: I understand, but I don't want Linux to became new Internet Explorer :P
<dos1> s/became/become/
<qi-bot> dos1 meant: "whitequark: I understand, but I don't want Linux to become new Internet Explorer :P"
<dos1> I'm not against custom extentions, new features outside of standard etc.
<dos1> I'm against IE mentality when doing so
<whitequark> well, IE was deliberately made incompatible
<dos1> and systemd story somehow triggers some warning light in my head
<dos1> hopefully it's false alarm :P
<larsc> well you can't run the BSD init system or the solaris init system or the MacOSX init system on Linux either, but I get your point
<dos1> I don't know much about non-Linux init systems - so maybe if I did, then the same light would blink as crazy when thinking about them ;)
<whitequark> sstemd looks quite a bit like launchd
<whitequark> (osx init)
<larsc> some features were inspired by launchd
<dos1> maybe except OS X, as I simply don't care
<dos1> I don't expect anything from Apple, as they can't expect me to be their customer ;]
<CYB3R> Anyone knows if Jolla is using systemd or not?
<whitequark> it makes no sense to be ignorant about design, even if you don't use the system just because of licensing terms
<dos1> whitequark: maybe, but if I won't use it because of licensing terms, then I don't care about it being portable and for sure never will
<larsc> it's a bit sad that the Unices and Unix-like OSs are drifting apart API wise. But systemd is not the problem
<larsc> merly a symptom of the problem if you will so
<whitequark> dos1: portability != design. I may never run OSX but I sure will look into how its kernel is done
<whitequark> (well, except its kernel is a decade behind Linux so I actually won't. but I could.)
<dos1> whitequark: :nod:
<larsc> "good artists copy, great artists steal" ;)
FDCX_ has joined #qi-hardware
FDCX has quit [Ping timeout: 245 seconds]
arielenter has quit [Ping timeout: 260 seconds]
arielenter has joined #qi-hardware
rz2k has quit [Read error: Connection reset by peer]
arielenter has quit [Ping timeout: 248 seconds]
arielenter has joined #qi-hardware
lekernel has quit [Quit: Leaving]
arielenter has quit [Quit: Leaving.]
jekhor_ has quit [Ping timeout: 250 seconds]
<wpwrak> (posix) that discussion is bizarre. posix was one of the biggest steps forward in the whole history of unix. before, there was a gazillion of diverging unices. that's was the situation that got us things like autocrap.
<nicksydney> wpwrak: was reading the mailing list about nano....when the project was started the ingenic chip was sourced from ingenic directly ?