Impresions new image 20091216

Wolfgang Spraul wolfgang at sharism.cc
Thu Dec 17 13:25:10 EST 2009


David,
spent a day with the great guys at our factory Sunty in Shenzhen today, a quite
productive day all about quality and making our devices better and learning
how we can make things smoother and cheaper and 100% GPL :-)

Saw this neat hinge torturing device in one corner, took a 23s video just
for fun (pretty strange video, I admit, but I liked it ;-))
http://downloads.qi-hardware.com/people/wolfgang/sunty_shenzhen_18dec2009/hinge_endurance_testing.avi

More importantly, everybody works really hard on the quality. Here is an
example of about 50 Ben NanoNotes sorted out for not meeting mechanical
quality criteria (see the little red arrow labels)
http://downloads.qi-hardware.com/people/wolfgang/sunty_shenzhen_18dec2009/nanonote_mechanical_yield.jpg

I flashed the new (and first) 1216 image to 40 Nanos to try to track down the
CRC boot bug. I will see whether I can boot all of them over the next couple
days. If I find one that doesn't boot anymore, maybe we have a chance to
hunt down the bug.
While doing that we noticed one device that showed a bug caused by the beads
added for CE compliance
http://downloads.qi-hardware.com/people/wolfgang/sunty_shenzhen_18dec2009/nanonote_beads_stripes.jpg
We will improve our testing procedures and require higher precision for the
beads - it was a good catch today.

Today I basically learnt the following things:

1. We have to invest more time in faster and easier flashing procedures. This
is good for end users, developers and production alike!
Especially the speed will be super important as we increase the NAND next year,
maybe to 16 GB or more, and want to fill it up with valuable free content,
and maybe even offer multiple images in our shop etc. All of this will only
be possible if we improve software in this area.
Some things we could do is reflashing from microSD (maybe another keypress
option in u-boot), reflashing via u-boot's DFU protocol like on the
FreeRunner, offering the usbboot utility in both i386 and x86_64 variants,
testing and supporting parallel flashing of multiple devices in usbboot,
(or dfu-util) - first 2, then more.

2. We need to hurry up the idea of working on free (GPL'ed) production testing
software. Good production testing software is the key to keep the quality high,
costs low, and also core to our idea of making the device easy to improve
and manufacture for others.

> distortion on speaker  on ben nanonote project
> added comment on power off issue

Thank you so much. Perfect! Same as Don Knuth we should offer some rewards for
finding bugs :-)
http://en.wikipedia.org/wiki/Knuth_reward_check
Any ideas what we could offer?

> Zimreader issue on openwrt-packages
> vido isue on vido project

Hmm. I think we should write them all up in the ben-nanonote project. Bugs is
all about visibility. So I think we should all point and look to one unified
place for a while. If that place gets too crowded, and some people just don't
want to be bothered with noise, it's better to branch out. But we are not there
yet. There are 4 issues written up in Ben NanoNote now (2 from you, thanks a
lot!). I will start writing up more too.
Do you agree with the location? Should we move the 2 issues from openwrt-packages
and vido over to ben-nanonote? I would even suggest to close the 'Issues' section
in openwrt-packages and vido?

> I have doubts on where to report the gtkperf isues, on ben-nanonote?

I'd say so, yes. Let's see what others say.

> All soft installed by default must work, if some soft doen't work out
> of the box please don't include it,

Absolutely. That will be exactly what we will do. Just understand the 20091216
image is the very first thing we can even consider an image because it meets
certain base criteria:
1. version number in /etc/VERSION to make bugs reproducible
2. all files needed to reproduce the image uploaded and archived on downloads
3. patent free to the best of our knowledge (no MP3)
It contains 97 packages, quite a few of them currently do _NOT_ meet your above
standard. In the coming weeks and months we will work towards this goal, with
the focus definitely on first getting what's in the image working well, before
adding new stuff.

> Agree, but please don't consider the only blocking factor to mark the
> release as production ready

Production will proceed nonetheless, this is a tough decision but as I've
explained before our manufacturer fully supports our whole open-sourcing thing
so we can re-use many closed components such as production testing software,
hardware design, etc. while we are slowly increasing the free'd (GPL'ed) parts.
Actually our manufacturer is quite excited about the project, they already
understand that if we are successful, they can 'take' our stuff and produce
whatever they like as well.
Thank you Google! (really, Android is helping a lot...)
So production & software progress are overlapping. I feel we have a
responsibility to ship as early as possible, in order to increase feedback and
grow the community. As long as we communicate clearly where the actual software
USABILITY stands that's the right approach I believe.

Nobody from Qi will make any false claims as to the software usability for an
end user: 20091216 is pretty much totally useless for any normal end user!
It's just our first real 'image' that stands on it's own feet (see above).
Do you agree?
Let's hope that 201001xx or 201002yy start to become valuable to normal
people. It's the wish of all of us to get there asap, we believe with free
software we can do it.

> Easy updates, to no need to reflash on each release. or if reflash is
> unavoidable include  a backup/restore data/packages installed from
> repos after reflashing in the reflashing script.

Also, full ACK. Easy updates are very high on our priority list.

> Some stetical  work has to be done(boot up logo,  to make first boot
...
> is a gtk-launcher in progress I remember, If there is some code to
...
> Easy (even more) reflashing (aka ease USB-boot, on this release
> power+U works, GOOD!! but to leave that state you have to  disconnect
...
> I will hear your opinions on this before put any ticket on track system

I suggest just write it up in ben-nanonote, IMO the best way to aggregate
feedback and process the proposals.
I like those _LOOONG_ Debian bug reports where people chime in over months
until a really good consensus and solution is found :-)

Thanks so much for your feedback, very encouraging!
Wolfgang

On Thu, Dec 17, 2009 at 11:48:27AM +0100, David Reyes Samblas Martinez wrote:
> 2009/12/17 Wolfgang Spraul <wolfgang at sharism.cc>:
> > David,
> > thanks so much for testing!
> I fun to play with such lovely device :) , so thank you(all) for make
> it to make me able to do this test :P
> <snip>
> > So we can start writing up bugs, then reproduce and fix them!
> > Maybe you can write up some of the things you found at
> > http://projects.qi-hardware.com/index.php/p/ben-nanonote/issues/
> Done,
> distortion on speaker  on ben nanonote project
> added comment on power off issue
> Zimreader issue on openwrt-packages
> vido isue on vido project
> 
> I have doubts on where to report the gtkperf isues, on ben-nanonote?
> <snip>
> > So the 'kernel CRC' error is my highest priority problem right now in terms
> > of declaring any image fit for use in production.
> Agree, but please don't consider the only blocking factor to mark the
> release as production ready
> I beleveve there are some other concerns that had to be considered to
> qualify software ready to final user.
> 
> All soft installed by default must work, if some soft doen't work out
> of the box please don't include it,
> 
> Easy updates, to no need to reflash on each release. or if reflash is
> unavoidable include  a backup/restore data/packages installed from
> repos after reflashing in the reflashing script.
> 
> Some stetical  work has to be done(boot up logo,  to make first boot
> experience sweet for non hacker costumers (the wider audience)  there
> is a gtk-launcher in progress I remember, If there is some code to
> work on I will like to take a look and help in whatever possible,
> Please hackers, dont start a flame war here, I love console too and
> surely in some case uses I will disable this launcher to not start up
> on boot, but a graphical frontend is needed to make it eyecandy to non
> hackers, advanced users, or whatever name you give to people using
> computers in a proper way :P
> 
> I consider this two ones a must to launch product to market, the
> following is a quite good enhancement to at least write it down
> 
> Easy (even more) reflashing (aka ease USB-boot, on this release
> power+U works, GOOD!! but to leave that state you have to  disconnect
> and remove battery, is technically possible to restart device using
> reset, power, or key combination?)
> 
> I will hear your opinions on this before put any ticket on track
> sistem, wiki or whatever place you  think it has to be writed (or not)
> 
> > That or anything else we find, let's please write it up in our bug tracking.
> > Thanks again for your testing, way to go!
> Again thanks for the fun :)
> > Wolfgang
> >
> > On Thu, Dec 17, 2009 at 12:32:37AM +0100, David Reyes Samblas Martinez wrote:
> >> Here is a brief report image flashe using the reflash_ben.sh :
> >>
> >> *Perceive Increased boot speed
> >>
> >> *Bigger fonts, personaly I prefer the previous ones but it it matter
> >> of taste I guess
> >>
> >> *Powerbutton don't work to shutdown
> >>
> >> *Sound still disable by default, you have to run alsamixer to enable
> >> Output Mixer DAC
> >>
> >> *Sound is too loud to the speaker, at 100 speaker distorts like a
> >> Manowar bass solo, I think actual 0% has to be the 100% level , on
> >> headphones volume is ok no distorsion, and sound is good.
> >>
> >> *Yes it can play wav files with aplay :) and sounds pretty good to
> >> such small speaker and low price device
> >>
> >> *ZimReader , I use this file that works on my PC[1] (3,2Gb) Program
> >> starts and I can connect to Ben 8080 port  on PC navigato it only
> >> loads the navigation part and was unable to serve an article, here is
> >> the output:
> >>
> >> #ZimReader -l 192.168.1.1 wikipedia_es_kiwix_120000\+_06_2009_alpha4.zim
> >> IP 192.168.1.1 port 8080
> >> Wikipedia ist jetzt unter http://localhost:8080/ verfügbar
> >> Die Einstellungen können unter $HOME/.ZimReader geändert werden
> >> ---[and after a while]-----
> >> 1969-09-25 10:23:17.54669 [663.7176] FATAL tntnet.worker - requesttime
> >> 600 seconds in thread 2051 exceeded - exit process
> >>
> >> also here is the top info when ZimReader is loaded
> >> Mem: 27380K used, 1092K free, 0K shrd, 36K buff, 14024K cached
> >> CPU:   0% usr   0% sys   0% nic   0% idle   0% io 100% irq   0% sirq
> >> Load average: 1.00 1.00 0.92 1/41 684
> >>   PID  PPID USER     STAT   VSZ %MEM %CPU COMMAND
> >>   654   647 root     R     1364   5%   1% top
> >>   676   627 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   677   676 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   684   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   683   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   679   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   680   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   681   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   678   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   682   677 root     S     9080  32%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   658     1 root     D     4692  16%   0% ZimReader -l 192.168.1.1 wikipedia_es
> >>   432     1 root     S     1372   5%   0% /bin/ash --login
> >>   440     1 root     S     1372   5%   0% syslogd -C16
> >>
> >> Too much memory hungry? maybe a swapfile can help?
> >>
> >> * vido, just a blank screen and I need to reboot the  device,  here is
> >> the output:
> >> root at BenNanoNote:/card/zipfiles# vido wikipedia_es_kiwix_120000\+_06_2009_alpha4
> >> .zim
> >> commandline read: vido
> >>
> >>    ~~~~~~~~~~~~~~~~~~~~~~~~~~| DirectFB 1.4.2 |~~~~~~~~~~~~~~~~~~~~~~~~~~
> >>         (c) 2001-2009  The world wide DirectFB Open Source Community
> >>         (c) 2000-2004  Convergence (integrated media) GmbH
> >>       ----------------------------------------------------------------
> >>
> >> (*) DirectFB/Core: Single Application Core. (2009-12-15 22:50)
> >> (*) Direct/Thread: Started 'VT Switcher' (624) [CRITICAL OTHER/OTHER
> >> 0/0] <2093056>...
> >> (*) Direct/Thread: Started 'VT Flusher' (625) [DEFAULT OTHER/OTHER
> >> 0/0] <2093056>...
> >> (*) DirectFB/FBDev: Found 'JZ4740 FB' (ID 0) with frame buffer at
> >> 0x01980000, 300k (MMIO 0x13050000, 64k)
> >> (*) Direct/Thread: Started 'Linux Input' (626) [INPUT OTHER/OTHER 0/0]
> >> <2093056>...
> >> (*) DirectFB/Input: gpio-keys (1) 0.1 (directfb.org)
> >> (*) Direct/Thread: Started 'Linux Input' (627) [INPUT OTHER/OTHER 0/0]
> >> <2093056>...
> >> (*) DirectFB/Input: matrix-keypad (2) 0.1 (directfb.org)
> >> (*) DirectFB/Graphics: Generic Software Rasterizer 0.6 (directfb.org)
> >> (*) DirectFB/Core/WM: Default 0.3 (directfb.org)
> >> (*) Direct/Thread: Started 'EventBufferFeed' (628) [MESSAGING
> >> OTHER/OTHER 0/0] <2093056>...
> >> GConf Error: Failed to contact configuration server; some possible
> >> causes are that you need to enable TCP/IP networking for ORBit, or you
> >> have stale NFS locks due to a system crash. See
> >> http://www.gnome.org/projects/gconf/ for information. (Details -  1:
> >> Failed to get connection to session: dbus-launch failed to autolaunch
> >> D-Bus session: Autolaunch requested, but X11 support not compiled in.
> >> Cannot continue.
> >> )
> >> GConf Error: Failed to contact configuration server; some possible
> >> causes are that you need to enable TCP/IP networking for ORBit, or you
> >> have stale NFS locks due to a system crash. See
> >> http://www.gnome.org/projects/gconf/ for information. (Details -  1:
> >> Failed to get connection to session: dbus-launch failed to autolaunch
> >> D-Bus session: Autolaunch requested, but X11 support not compiled in.
> >> Cannot continue.
> >> )
> >>
> >> gtkhtml-ERROR **: gconf error: Failed to contact configuration server;
> >> some possible causes are that you need to enable TCP/IP networking for
> >> ORBit, or you have stale NFS locks due to a system crash. See
> >> http://www.gnome.org/projects/gconf/ for information. (Details -  1:
> >> Failed to get connection to session: dbus-launch failed to autolaunch
> >> D-Bus session: Autolaunch requested, but X11 support not compiled in.
> >> Cannot continue.
> >> )
> >>
> >> aborting...
> >> (!) [  622:    0.000] --> Caught signal 6 (sent by pid 622, uid 0) <--
> >>
> >> *gtkperf, scrolling on gtkTexview text box and in firt tab test
> >> results box produce is slow and produce a black box than hide text but
> >> on Help tab text box scroll works flawlessly , Drawing doesn't draws
> >> anything
> >>
> >> Well that's all for now, do you want I test anything more on the just
> >> reflashed distro?
> >>
> >> Keep on the good work :)
> >>
> >> David Reyes Samblas Martinez
> >> http://www.tuxbrain.com
> >> Open ultraportable & embedded solutions
> >> Qi inside NanoNote, Arduino, Openmoko, Openpandora
> >> Hey, watch out!!! There's a linux in your pocket!!!
> >>
> >> _______________________________________________
> >> Qi Developer Mailing List
> >> Mail to list (members only): developer at lists.qi-hardware.com
> >> Subscribe or Unsubscribe: http://lists.qi-hardware.com/cgi-bin/mailman/listinfo/developer
> >
> > _______________________________________________
> > Qi Developer Mailing List
> > Mail to list (members only): developer at lists.qi-hardware.com
> > Subscribe or Unsubscribe: http://lists.qi-hardware.com/cgi-bin/mailman/listinfo/developer
> >
> 
> 
> 
> -- 
> David Reyes Samblas Martinez
> http://www.tuxbrain.com
> Open ultraportable & embedded solutions
> Qi inside NanoNote, Arduino, Openmoko, Openpandora
> Hey, watch out!!! There's a linux in your pocket!!!
> 
> _______________________________________________
> Qi Developer Mailing List
> Mail to list (members only): developer at lists.qi-hardware.com
> Subscribe or Unsubscribe: http://lists.qi-hardware.com/cgi-bin/mailman/listinfo/developer




More information about the discussion mailing list


interactive