<kristianpaul>
por siero para SIE estaba pensado en una vesio modificada de un bash o algo, que de un menu con opciones
<kristianpaul>
todo esto a falta de no tener teclado pero si una pantalla ..
<kristianpaul>
ok
<rafa>
kristianpaul: (tenes que tener para SDL los paquetes libsdl-1.2-0 y libsdl-1.2-0-dev instalados)
<kristianpaul>
ya ya
<rafa>
libsdl-1.2-0 seguramente ya esta instalado por default
<kristianpaul>
sip
<kristianpaul>
pero los intente instalar y lo note !
<rafa>
:)
<kristianpaul>
nite
<kristianpaul>
and see you inn few hours
<rafa>
nite!
<viric>
kristianpaul: el sie no tiene ningún teclado?
<kristianpaul>
viric: no
<viric>
vaya.
<kristianpaul>
tiene IO por todo lado pero ninguna tecla arra o abajo y enter :p
<viric>
a lo mejor se puede acoplar aunque sea un ps/2
<kristianpaul>
ee buena idea
<kristianpaul>
ps/2 !
<viric>
es muy fácil el protocol
<viric>
o
<kristianpaul>
sip
<viric>
no se si va a 12V como rs232, pero.
<kristianpaul>
no creo es ttl
<kristianpaul>
pero no se si ttl de 3v3
<viric>
no creo que de 3v3
<kristianpaul>
:/
<rafa>
xiangfu: hi, is the nand partition defined into kernel right?
<viric>
rafa: you can override it by the kernel command line
<viric>
(which is hardcoded in uboot. hehe)
<rafa>
viric: nice :P..
<viric>
check /proc/cmdline.
<viric>
to see if it is uboot setting it
<rafa>
viric: before the last change in qi-openwrt people was able to boot debian/jlime from uSD and mount the openwert on nand, but no now after the new layout (256MB for rootfs, rest for data)
<xiangfu>
rafa: yes. it's hardcoded in kernel.
<rafa>
xiangfu: larsc : any idea what to do?.. before we mounted okey, with the new layout no
<viric>
how can they boot from uSD?
<viric>
does uboot try to boot from uSD first?
<viric>
rafa: you would need the debian kernel to use the same layout as used in the system in nand
<rafa>
viric: some versions of uboot lets you press [M] and power butoon
<viric>
that means 'boot from sd'?
<rafa>
yes.. I meant sd
<viric>
but it may work only with one kind of filesystem in uSD, and the kernel at some specific place
<viric>
I could read the uboot source to know the details...
<viric>
rafa: so, I don't know if the nand partitions can be set after the kernel booted
<rafa>
viric: why debian kernel?
<viric>
rafa: does the debian run with the kernel in the nand?
<xiangfu>
rafa:Â Â not boot from uSD or not mount the NAND rootfs ? or both not work?
<rafa>
viric: you are the guy who answers with another question ? ;)
<xiangfu>
rafa: boot from SD should works fine. because we only change the NAND partition.
<rafa>
xiangfu: yes, sorry. I will try to explain better:
<rafa>
xiangfu: before the last change.. we were able to boot from SD and then mount the nand under /openwrt for example. Now no
<viric>
rafa: it was me who was surprised that the nanonote can boot from sd :)
<xiangfu>
viric: yes. nanonote can boot from sd :)
<rafa>
viric: sorry, I am not sure the current state of debian. It worked before I would say. I am talking about "other" in general, and specifically I am talking for Jlime
<viric>
ahh ok
<viric>
so, the jlime kernel should boot with the proper nand partitioning
<viric>
which maybe can be set at runtime, but I only know it can be set at the kernel cmdline
<rafa>
viric: mmh?.. I do not understand. jlime kernel boot from sd and from nand.. you can install jlime on nand or sd.. That is okey and all works
<viric>
ah really? somehow it  has to know the proper nand partitioning
<rafa>
viric: but if you boot, for example, jlime from sd.. and you also have openwrt on nand, and then you want to access your files on openwrt from jlime you can not mount that nand right now
<viric>
rafa: maybe jlime boots fine from nand because it sets its own partitioning (different from openwrt)
<rafa>
viric: jlime has all the files and docs to install it on nand or sd.. you could try ;)
<viric>
rafa: ok :)
<rafa>
viric: yes, sure.. but jlime does not use the openwrt files if that is that you meant.. when you install jlime on nand you write a new rootfs on nand..
<rafa>
the jlime rootfs.
<rafa>
and yes, jlime kernel knows its own layout
<viric>
exactly.
<rafa>
viric: but jlime kernel is qi kernel.. just that it was built a time ago
<viric>
and it does not coincide with that of openwrt
<viric>
well, then you already have the problem :)
<rafa>
viric: and we are not going to change the nand layout if there is not a good explanation why it would be the best..
<rafa>
viric: me?.. why.. now you find jlime only.. but what if there are 5 distributions around?.. all of them should change those layouts when sombedy changes it ? :=
<rafa>
:)
<rafa>
that does not sound as a good idea
<viric>
why not go setting uboot to determine the nand layout?
<viric>
then all kernels would coincide
<viric>
oth, it would be difficult to override without overwriting uboot instead of only the kernel
<rafa>
viric: mmmh?.. into uboot we woult have different layouts for different linuxs?
<viric>
rafa: no, same layouts for all linuces
<viric>
rafa: but if we can find how to reset the mtd layout at runtime instead of boot time, you can run a script to set that to whatever you want
<rafa>
viric: we had the same layout for all linuces.. but now qi - openwrt changed its own.. and now we have different layouts. And what about if you do not like the idea 256MB for rootfs and rest for data?
<viric>
rafa: you build your own uboot with whatever you want
<rafa>
viric: no.. that is not a good idea
<viric>
in NAND, there is no 'partition table', so...
<rafa>
viric: I mean.. we are trying to be compatible with qi openwrt (we=jlime).. no comes here with those ideas.. like "you can built a new..."
<rafa>
viric: just try to be okey with all I would say
<viric>
jlime can't be easily compatible if openwrt does not want to ;)
<rafa>
viric:Â Â that is not the idea what we want to have.. we have been talking about dual boot between qi and jlime
<rafa>
viric: but well, now we have another piece what either qi needs to change or jlime needs to change
<rafa>
viric: that is my concern
<rafa>
viric: I mean.. I do not care.. I am talking because I would like to help with that possibility (users using both: jlime and openwrt), just for fun for all.
<rafa>
viric: I just would like to know if somebody was thinking about how to work that change with others before to do that.
<viric>
ok, clear
<xiangfu>
which commit you use for compile the last kernel?
<rafa>
xiangfu: current jlime released versions use 2.6.34 kernel, I can tell you if you need to know the commit. Current jlime under development uses latest 2.6.36 kernel and qi patches.
<bartbes>
so is libogg/vorbis/vorbisfile slow on the nn?
<bartbes>
because alsa keeps telling me I have buffer underruns (which i can hear, obviously)
<xiangfu>
rafa: ok,  hmm.. let me search about the command "ubiattach",  I need to know how "ubiattach" get the partition. or if we can set the partition to "ubiattach"
<rafa>
xiangfu: ah.. sounds as good idea if that works
<wejp>
bartbes, you should not use libvorbis on the nanonote. you will never get an acceptable speed with that
<wpwrak>
i was wondering how long it would take until someone would complain about the rootfs size change :-)
<B_Lizzard>
Is anyone interested in testing a preview of the newer Muffinman image?
<kyak>
B_Lizzard: when i choose "End Session", how do i make the dropdown list appear?
<bartbes>
wejp: and that thanks to you ;)
<B_Lizzard>
kyak, space
<xiangfu>
kyak: my daily build on build host is start. so I will test the gcc-mips on build hosts tomorrow. then re-enable it in config.full_system :)
<xiangfu>
s/start/started
<kyak>
B_Lizzard: k, thnkas!
<B_Lizzard>
larsc, one more thing I forgot to mention, usb0 doesn't appear to exist in 2.6.36.
<kyak>
xiangfu: that's great.. and i think new jlime will inspire us for more.. i see a very interesting SDL puzzle game there
<B_Lizzard>
That's a port of stppc2x, which is originally for the GP2X
<B_Lizzard>
I could've used the normal puzzle games in GTK2 but that's slow as molasses
<kristianpaul>
larsc: i sadly confirm no usb0 :'(
<larsc>
kristianpaul: the cpu resumes when an interrupt occurs
<kristianpaul>
larsc: k
<kyak>
B_Lizzard: i assume the network is not working?
<kristianpaul>
is not kyak
<B_Lizzard>
kyak, yes.
<B_Lizzard>
Um, no
<kristianpaul>
no :'(
<larsc>
the usb gadget driver for the jz4740 is not in upstream yet.
<kristianpaul>
ahh
<larsc>
so you need to use the one from the qi-hardware tree
<B_Lizzard>
Ah, thanks.
<B_Lizzard>
I probably forgot that one.
<kristianpaul>
okay boots in 14 seconds
<kristianpaul>
including uboo
<kristianpaul>
t
<kristianpaul>
larsc: are you aware of power consuption in suspend mode?
<larsc>
8 ma or something on the nanonote
<larsc>
mA
<kristianpaul>
he
<kristianpaul>
thats a long time in my pocket
<B_Lizzard>
Is everything suspended apart from the keyboard?
<B_Lizzard>
I guess there's not much to suspend but ehm.
<kristianpaul>
yeah
<kristianpaul>
LCM
<kristianpaul>
cpu !!
<kristianpaul>
is not enought ??
<B_Lizzard>
:)
<kristianpaul>
larsc: interrupt source could also be one of the other gpio pins or just the keyboard ones?
<larsc>
well there are other jz4740 devices where it gets below 1 mA
<kristianpaul>
ohh
<larsc>
kristianpaul: k kristianpaul  ny interrupt can be used as a wakeup source
<larsc>
kristianpaul: any interrupt can be used as a wakeup source
<kristianpaul>
ok
<kristianpaul>
larsc: RTC is a interrups source?
<larsc>
you just have to tell the kernel that you want to use it as one
<larsc>
kristianpaul: yes
<kristianpaul>
onw i'm wondering if it could work as alarm clock !!
<rafa>
xiangfu: how long it takes for openwrt to boot?
<kyak>
B_Lizzard: what about other apps that used to be in jlime? like prboom, mydict, organizer or what else.. or they are just not there is test image?
<rafa>
xiangfu: until GUI is ready to use for users
<B_Lizzard>
kyak, this image and rafa's images are different
<B_Lizzard>
I plan on implementing stuff but I want the base system to mature
<B_Lizzard>
Ugh
<larsc>
kristianpaul: it can. Ayla wrote a patch for it.
<kristianpaul>
if uboot is eliminated from this history boot time could achive around 10 seconds i guess
<B_Lizzard>
In any case, most applications aren't there because this is a preview
<kristianpaul>
but are there
<kristianpaul>
how is that?
<kristianpaul>
ahh you meant common apps?
<B_Lizzard>
Yeap
<xiangfu>
rafa: it's need ~20 secs until the gmenu2x show up. ~3 secs the OpenWrt slash screen show up.
<wpwrak>
grmbl. PxFLGC == PxDATS ? i don't think so ... now, where do you send bug reports for documents you nominally don't have access to ? :)
<larsc>
wpwrak: actually it is
<wpwrak>
larsc: *hmm* looks like a very odd choice then. can't this cause conflicts ?
<wpwrak>
oh, it gets better :-) "They are read-only registers" ... "when write 1 to the bit, [...]"
<kristianpaul>
hw bug?
<wpwrak>
kristianpaul: the address equality or the "read-only" vs. "write" ?
<kristianpaul>
the "[...]"
<kristianpaul>
i could imagine lots of thing after you tried write it
<wpwrak>
kristianpaul: oh, i just omitted it. the contradiction is in calling it first read-only (which it isn't) but then explaining what happens when you write it.
<wpwrak>
kristianpaul: this is simply a documentation bug.
<kristianpaul>
ah ok :)
<kristianpaul>
"be prapared to documentation lie to your face"
<kristianpaul>
wee my new soerkis based server booted :)
<wpwrak>
(lying documentation) yeah :)
<wpwrak>
what's odd on the hardware side is that the (interrupt) flag clear register would occupy the same location at the data set register.
<wpwrak>
larsc: (conflicts) hmm, apparently not. i see that a pin can't be GPIO and interrupt at the same time. also a bit of an odd choice.
<larsc>
well, it does not makes sense to have a gpio in output and interupt mode at the same time
<larsc>
on the other hand you can still read the input values when the gpio is in interrupt mode
<wpwrak>
but why would there have to be a distinction between gpio and interrupt mode in the first place ? sure, it would be useless in most cases, but there's probably a signal path to the interrupt logic anyway, so why introduce a mode ? if you don't care about the pin, you could just mask it.
<larsc>
well there might be reasons why its done the way its done. but as long we dont know what the acutal hw implementation looks like, we can only speculate about it
<wpwrak>
yeah, this is true. pity, would be nice to have a little more orthogonality.
<rafa>
We need a qi / reseller server to do a jlime repo mirror without packages with patented technologies. Is tuxbrain or qi interested?
<kristianpaul>
rafa: *i think* that fidelio dot qi.com can host that but who is qi??
<kristianpaul>
rafa: ask to the list
<wpwrak>
kristianpaul: are you sure qi.com has anything to do with qi-hardware ?
<wpwrak>
kristianpaul: there is also no host called fidelio.qi.con
<wpwrak>
err, .coM
<kristianpaul>
wpwrak: not sure..
<kristianpaul>
well with qi.com i meant qi-hardware.com :p
<kristianpaul>
s/qi.com/qi-hw.com
<wpwrak>
there is also no host fidelio.qi-hw.com :-)
<kristianpaul>
ahh
<kristianpaul>
well you
<kristianpaul>
s/qi-hw.co/qi-hardware.com
<kristianpaul>
know
<wpwrak>
anyway, qi-hardware.com sounds like the right place. i guess wolfgang would handle decisions regarding such a mirror himself
<kristianpaul>
indeed
<rafa>
wpwrak: yes.. but where is that lazy guy? (wolf).. argh.. well, you have better work timezone than me to find wolfgang.. We need to ask him then
<rafa>
about lazy: of course, I am kidding :)
<wpwrak>
rafa: maybe a question for the openwrt gurus :-)
<rafa>
wpwrak: yeah :)
<wpwrak>
rafa: everybody here seems asleep. lazy bastards :)
<wpwrak>
rafa: does SDL take care of the ben's unusual pixel format ? or does the application have to compensate for it ?
<FrankBlues>
waves
<rafa>
wpwrak: about ipk. ipkg home page says about ipkg format: "In keeping with the "Just like Debian" philosphy laid out above, I first made the .ipk format equivalent to .deb, (but renamed to avoid frustration and confusion). However, I recently changed it to be a tar.gz file rather than an ar archive so that ar would not have to be installed to use ipkg. (But with 0.99 we changed it back because gzipping compressed data is a waste of CPU and battery)."
<rafa>
"(But with 0.99 we changed it back because gzipping compressed data is a waste of CPU and  battery)."
<wpwrak>
waves back :)
<rafa>
then.. a few lines after it says: "So, a .ipk file is an archive (either tar or gzipped tar) containing 3 members: "
<rafa>
nice.. who knows exactly..
<wpwrak>
rafa: so that's ... 4 format changes and maybe one name change. in what time interval ? :-)
<FrankBlues>
rafa: Just installed the Muffinman beta 4 - nice!
<rafa>
NO idea :).. OpenEmbedded does ipk, opk, or deb I would say.. but all of them are ar archives
<rafa>
wpwrak: (that was for you)
<rafa>
FrankBlues: he.. great! :)
<rafa>
FrankBlues: if you like it to use every day please read the Muffinman documentation wiki page.. it is just one page :) ..and it has a lot of useful information and lot of tips
<wpwrak>
rafa: perhaps the best approach is to just  for n in /bin/* /usr/bin/*; do for m in "" x -x xf xfz xfj; do $n $m $package && return; done
<wpwrak>
rafa: i.e., see if *anything* can handle the file ;-)
<rafa>
wpwrak: SDL: sorry, what do you mean with unusual pixel format?
<wpwrak>
rafa: the ben's display has the RGB sub-pixel is a different location than "normal" lcd displays for computers. that was for instance a problem for text fonts. perhaps you remember this.
<wpwrak>
rafa: hmm, is there a basic and common library with the usual 2D graphics primitives for SDL ? e.g., drawing a line, a filled rectangle (or polygon), etc.
<rafa>
yes I remember, but I would say that SDL does not do anything about. At least the libSDL that OE has it the standar one. If that problem is a known problems in many machines and sdl can controls that I would be impressed :).. Well, I have not seen this problem before and how to fix that
<wpwrak>
(pixel problem) hmm, so you say it doesn't happen in a noticeable way ? or that it happens in a noticeable way, but nobody cares to fix it ?
<rafa>
wpwrak: I would say that it happens but nobody cares.. But I am not sure.. every sdl game I tested looked nice :)
<wpwrak>
(pixel problem) okay. i'll see how this goes then :)
<rafa>
wpwrak: SDL.. maybe libsdl-sge is good for primitives.. but it is not on jlime repo I see :( .. with standar sdl you just can do rectangles.. or a pixel.. also you can load graphics.. so you can load triangles as surfaces or other kind of primitives.. and to put them.. But well, if you want to use different sizes or geometry you need another lib like libsdl-sge
<wpwrak>
any suggestion for a drawing lib ? i see that there are a bunch of them, but which would be the "most standard" ?
<wpwrak>
ah, messages crossed :)
<rafa>
:)
<rafa>
anyway, libsdl-sge should not be hard to port
<rafa>
wpwrak: that would be the best way to do graphics on Ben.. SDL using x11 or fb is fast.
<rafa>
bah.. wolfgang is not logged in yet.. I go to sleep.. cya
<wpwrak>
(sdl_gfx) looks nice. and it has the functions i'm looking for. thanks !
<wpwrak>
and of course, the openwrt host toolchain doesn't have sdl :-(
<rafa>
wpwrak: he.. murphy.. but well, sdl is the exact thing in pc version so you just init sdl as a 320x240 window on your pc and try there.. then the port is ultra easy and you do not need to touch anything in the code to build for ben
<rafa>
exact=exact same
<wpwrak>
rafa: (easy port) yup, that's what i'm hoping for :)