<kyak>
wolfspraul: hi! i just sent an e-mail to discussion list with attachment ~900Kb, but it was held due to 200Kb size limitation. Should i upload the attachment somewhere else and provide a link instead?
<wolfspraul>
kyak: that would be more in line with general mailing list etiquette, yes
<wolfspraul>
thanks for asking btw
<kyak>
i will keep that in mind
<kyak>
i'm not a power user of mailing lists :)
<kyak>
i will resend it then
<wolfspraul>
ok then I will discard the queued mail, so that nobody else approves it and we have it twice...
<kyak>
xiangfu: i think i found an issue in gmenu2x skins..
<kyak>
when you change the skin to "backup" (and save) and then change back to "Default", gmenu2x won't start after restart
<kyak>
i hunted it down to "messageBoxBg=#00080" line in /usr/share/gmenu2x/skins/Default/skin.conf.. this line is "messageBoxBg=#00000080" in default skin.conf
<kyak>
so seems like gmenu2x is not saving its settings correctly
<kyak>
so, after change "messageBoxBg=#00080" back to "messageBoxBg=#00000080", gmenu2x would start fine
<qi-bot>
[commit] Werner Almesberger: atben.brd: rearranged component references and values for printing/display http://qi-hw.com/p/ben-wpan/84278c8
<qi-bot>
[commit] Werner Almesberger: atben, atusb: set solder mask and solder paste clearances to NXP's values http://qi-hw.com/p/ben-wpan/daaac58
<xiangfu>
kyak: hmm... just grep the source code.  seems gmenu2x never using this "messageBoxBg" , there is one line, but it's in comment
<kyak>
xiangfu: there are two of them..
<xiangfu>
kyak: yes. one in comment, another in a static char*
<kyak>
if you continue digging, you'll also find COLOR_MESSAGE_BOX_BG
<kyak>
gmenu2x.cpp, line 99
<kyak>
test,test
<kyak>
i had > 100 s lag... thought i'd be disconnected :)
<kristianpaul>
ping ping kyak
<kristianpaul>
:-)
<kyak>
xiangfu: i wanted to ask your opinion on something. There are packages in openwrt feeds explicitely linked against libncurses (non-wide version): irssi, for instance. How should we do if we want to have all those apps linked..
<kyak>
..against wide ncurses?
<kyak>
kristianpaul: thanks ;)
<xiangfu>
kyak: recently I have one idea about packages. let me explain first. (maybe this idea can work on this issue and libiconv (stub version) )
<xiangfu>
let me start with libiconv-full libiconv-stub libiconv
<kyak>
oh good
<kyak>
i though about "conditional dependencies" mechanism, that is already there in openwrt
<kyak>
something like DEPENDS:=+CONFIG_PACKAGE_libncursesw:libncursesw
<kyak>
and then surround this "-lncurses" with if's (and link against lncursesw, if it's selected)
<xiangfu>
there is on "/usr/lib/libiconv-full" anymore. both libiconv-full libiconv-stub install to same place. and they have their names, then libiconv have one line in DEPENDS
<xiangfu>
just like you just wrote.
<xiangfu>
s/there is on/there is no/
<kyak>
i think this could work
<xiangfu>
I plan to test this tomorrow.
<kyak>
great!
<xiangfu>
today, I testing David's commit about plplot.
<xiangfu>
kyak: I guess our config.full_system have make rootfs > 512M :)
<kyak>
oh!!
<kyak>
i'm still living in 256 M :)
<kyak>
maybe we should consider and remove some rarely used packages.. build them is "modules"
<kyak>
*as
<xiangfu>
I have create a rootfs, (only without emacs), the ubi file is  ~ 429M.
<kristianpaul>
(i'm still living in 256 M ) oh, i could not agree better  :-)
<xiangfu>
after reflash it's using 470M
<xiangfu>
kyak: yes. agree, but we don't need build them as 'modules'. because we have 'config.all_packages' now.
<xiangfu>
now we can success build ~ 1700 packages.
<kyak>
xiangfu: when you build a release image, how do you know which packages to include withing the image and which to build as "modules"?
<xiangfu>
kyak: we always using "config.full_system", so all the packages in "config.full_system" will included,
<kyak>
xiangfu: ok, why do you need the config.all_packages then?
<xiangfu>
kyak: all_packages try to build all ~1700 packages.
<kyak>
does it mean that you build two times? first, you build config.all_packages and put them in downloads.qi-hardware.com/software/packages/NanoNote/Ben/latest/, and then you build config.full_system to make the release image?
<kyak>
yeah, is it empty because it is still building?
<xiangfu>
(this build start at : (21-Jan-2011 22:02) check the VERSIONS file, and today it's still building ...)
<kyak>
not good, i think you could bug wolfspraul about that
<xiangfu>
kyak: yes. you can see the BUILDLOG time changes every minute.
<kyak>
right
<xiangfu>
kyak: this time , we want release "mirror sources" "all packages" and "images"
<xiangfu>
"mirror sources" is the "dl" folder.
<xiangfu>
kyak: also maybe next time we can usinig 'opkg update' for update the rootfs. but first we need keep all files in package.
<xiangfu>
there are still some files under "data/qi_lb60/files"
<kyak>
yeah, right, very good idea
<kyak>
this could take a while
<kyak>
(upgrading via opkg)
<xiangfu>
kyak: yes. maybe we also need create packages for u-boot and kernel.
<kyak>
xiangfu: btw, did you get a feedback about building monolythic kernel?
<kyak>
i.e. not having sound as modules
<xiangfu>
kyak: someone reply email, said they using the monolythic  kernel for long time.
<kyak>
$ grep "Package:" tmp/.packageinfo |wc -l
<kyak>
3348
<kyak>
xiangfu: mean we only have around 50% packages build successfully?
<xiangfu>
kyak: ./scripts/feeds list | wc -l
<xiangfu>
2629
<kyak>
oh, much better :)
<xiangfu>
I want release images at Wednesday or Friday. so this time the sound will still build as module.
<kyak>
some packages are related to X, not to be counted.. and maybe some can't be built on mips
<kyak>
and some are patented..
<kyak>
xiangfu: when building config.all_packages, are there packages that fail to build?
<kristianpaul>
ad astersik (*) in hexdump mean i avoid show this lines because are the same as the one i already showed up to address just before the asterisk (*), isnt ?
<kristianpaul>
s/ad/an
<kristianpaul>
oh yes
<kristianpaul>
nv
<xiangfu>
kyak: for now. the php5 fail,
<xiangfu>
plplot-octave and plplot-tcl fail, but David have commit something, don't know if they are compile fine.
<xMff>
I also have commit access to upstream opkg in case some fixing is needed
<bartbes>
well then
<bartbes>
what are you waiting for? :P
<xMff>
spare time ;)
<Jay7>
sshfs :)
<kristianpaul>
wpwrak: you think "GPS-SDR I/Q data avaliable for Analysis" is a good subject for an update mail about the status of gps-sdr project for the ben nanonote?
<wpwrak>
sure. why not. any update is good :)
<kristianpaul>
k
<xiangfu>
kyak:  about all_pakcage and full_system, the better idea is move  "CONFIG_ALL=y" to config.full_system. then we can delete the all_packages.
<kyak>
xiangfu: then all packages will be built as "modules", and those that explicitely enabled (i.e. "=y") will be build into image?
<xiangfu>
kyak: yes.
<kyak>
this is good
<kyak>
we need to reduce the number of =y packages
<kyak>
>512 Mb image is so insane :)
<kristianpaul>
(finally some one listen to my toughts :-))
<kristianpaul>
kyak: can you write that to mail list, i think this deserves to be discussed
<kristianpaul>
s/list/list ?
<kyak>
kristianpaul: i really wouldn't like to start this discussion :) there will be many argues what needs to stay in image, and what needs to go away
<xiangfu>
kyak: yes. agree.
<kristianpaul>
well at least i tried to convince you ;-)
<xiangfu>
(I mean >512 Mb image is so insane :)
<kristianpaul>
Just be prepared for the next resize :-)
<qi-bot>
[commit] Werner Almesberger: 0805-6.fpd: changed from package-based definition to follow Johanson's design http://qi-hw.com/p/ben-wpan/1c573d2
<wpwrak>
while you wrote "Now with help from Fabrizio Tapper from CTAE"
<kristianpaul>
well, seems yes, i hope, but i just take his name from his mails
<kristianpaul>
well he is italian, so mistakes with last names are usual, like trying to write/pronounce  my last name after from a enlgish native speaker ;-)
<wpwrak>
;-))
<Ornotermes>
wpwrak: does atben only use SPI signals for control?
<rjeffries>
this.. 12:26 <Ornotermes> wpwrak: could be interesting to build small gadgets that could interact with ben/pc wireless
<rjeffries>
12:28 <wpwrak> Ornotermes: indeed :)
<rjeffries>
12:30 <Ornotermes> a small breakoutbord for uSD-card holder is all that's needed, no point in build another version of the board really
<rjeffries>
wprak what does the software for atusb look like? (before you add the 6LoWpan layers? just simple stream of bytes
<wpwrak>
rjeffries: the idea for atusb is to just pass the basic accesses to the chip. maybe combine some low-level functions where convenient. nothing more. the bulk of the stack would be on the host.
<wpwrak>
rjeffries: i.e., the atusb firmware wouldn't know how to send or receive a packet
<rjeffries>
ok fine. can it ship byes to radio to be transmitted and read the stream that comes in
<steve|m>
"usb -> spi interface" :)
<rjeffries>
s/byes/bytes
<rjeffries>
spi is on the radu chip correct
<rjeffries>
s/radu/radio/
<wpwrak>
steve|m: yup. a tiny bit higher-level, though. but not much.
<wpwrak>
btw, all the code is there, you can easily read what it does yourself ;-)
<wpwrak>
rjeffries: the transceiver has a few access commands: mainly registers and frame. i just map them to USB control transfers.
<rjeffries>
SCLK: Serial Clock (output from master);
<rjeffries>
MOSI; SIMO: Master Output, Slave Input (output from master);
<rjeffries>
MISO; SOMI: Master Input, Slave Output (output from slave);
<steve|m>
wpwrak: I'm planning something similiar for that m8c rfid reader.. have to write as least as possible in m8c asm :)
<wpwrak>
rjeffries: in the future, it may be convenient to combine some such commands into a single operation, though. but it would still be at the same level of abstraction.
<rjeffries>
so I wiull admit I do not have git installed so I think I can not read your source is that correct
<wpwrak>
rjeffries: but git helps. local browsing is always nicer than webbish interfaces
<rjeffries>
unferstood
<wpwrak>
the idea is to keep the device-specific code to a minimum (firmware and kernel driver). then go to some "generic" at86rf231 driver. and so on.
<roh>
sounds more like a task for something bigger than a 8bit cpu
<kristianpaul>
oh, MM1 RC/RUN 3 on the way...
<kristianpaul>
wpwrak: are you making a kernel driver for atben? i tought it was just bitbanging
<rjeffries>
does this mean a new softcpu is planned for MM? <kristianpaul> oh, MM1 RC/RUN 3 on the way...
<kristianpaul>
rjeffries: no no, just other batch of Milkymist One boards i guess :-)
<rjeffries>
i see
<rjeffries>
are any users using MM for VJ gigs or is it a little too rough aroung the edges for now (understandable)
<Joelixny>
Is this channel for electronic help like cuircuits and stuff? =)
<kristianpaul>
Joelixny: nope, try #electronics
<Joelixny>
ok thanks
<kristianpaul>
rjeffries: for softcore Milkymist questions i think proper place is #milkymist
<rjeffries>
I stad corrected
<rjeffries>
s/stad/stand/
<rjeffries>
MM is not relly on my radar yet. was just curious is all.
<kristianpaul>
sure :-)
<wolfspraul>
rjeffries: it's not too rough
<wolfspraul>
it works like a charm, out of the box
<rjeffries>
hi wolf. it is none of my beeswax as they used to say. it is a highly complex bundle of FRGAcode so I'd be totally amazed if there are not plenty of teething problems
<rjeffries>
s/FRGAcode/FPGA code/ next topic . <g>
<kristianpaul>
you should not care about FPGA actually. what Wolfgang points is true, i realized as one get my M1 and just pluged :-)
<wolfspraul>
I'm not known for exaggerating about tech features I think.
<kristianpaul>
and Milkydrop efects are wow, i got amazed watching/playing with some of that day
<wolfspraul>
there is one bug left, a power-up bug
<wolfspraul>
but it's rare and there is a workaround. We hope to track it down before rc3.
<rjeffries>
one last comment befor I apply duct tape to my lips: there is a difference between one bug you know of and all the ones nobody has found yet. I am DELIGHTED that the VJ app works, cool beans. I thought somebody had an issue with large enet frames.
<rjeffries>
back to my project to make an IR dongle for Nanonote. Wolf, no, at this stage I do not yet need a Nanonote
<rjeffries>
I have to learn how to make my own PCB (the 8:10 to breakout cable)
<rjeffries>
then with a loy of study and possinly a little help from Samrt People get the schematic done and reviewed
<rjeffries>
after I build one, borrow my friend's unused Nannote for a while
<rjeffries>
But first I need to convince him to man up and update to latest image
<kristianpaul>
In the meanwhile you can make your PCB and Schmatics :-)
<kristianpaul>
Yes, i'm having issues with some networking support on the RTEMS port for MM1, but that dint affect Flickernoise or Vj Stuff
<kristianpaul>
Is a requirement for my particular need.
<rjeffries>
kristianpaul what is your use case (curiosity only)
<kristianpaul>
I need transfer some files from the MM1 to my Desktop Computer using the ethernet
<kristianpaul>
or in some _theorical_ case, stream the data
<rjeffries>
ok. what's the a[[licayion or is this a secret that may earn you a gazillion Euros
<kristianpaul>
no no
<kristianpaul>
Did you read the last mail i sent to the list?