<kyak>
xiangfu: i ended up shrinking abook interface to fit to Ben display :) 4x8 font is just not readable if you use it long enough
<xiangfu>
kyak: great patch 001-term-size.patch.
<kyak>
there might be something left that can't fit into screen.. hope we will stumble upon it while using abook
<kyak>
btw, if we want utf-8 support in abook (and other ncurses apps), we would need libncursesw, which is not backported to backfire yet
<rafa>
wpwrak: interesting.. between packages there is an om-maps-berlin..ipk  but then .. I also saw an om-maps-buenos-aires..ipk  ;-))
<wpwrak>
yeah. and i never actually tried it. shame on me !
<rafa>
wpwrak: I just realized that .. so I never tried either
<rafa>
ah.. hong kong as well
<rafa>
london, paris, new york
<xiangfu>
kyak: maybe we can backport to openwrt-package.git,
<kyak>
xiangfu: good idea, wouldn't it cause conflicts with ncurses from packages/ ?
<xiangfu>
kyak: not sure, need test
<kyak>
i'm wondering how it'll work.. but this is definitely worth it.. also taking into account that though libncursesw is ported to the latest openwrt, it still doesn't work ;)
<kyak>
(well at least wide-char libncursesw "Hello world" didn't work for me)
<kyak>
so if have it in our packages, we can try and figure that out
<kyak>
i noticed that libncurses even from the latest openwrt is lacking a lot of patches
<kyak>
heh, it just appears twice in menuconfig
<xiangfu>
kyak: oh. you fast. I am reply Mirko's email.  about next openwrt release.
<kyak>
xiangfu: though it appears twice in menuconfig, the one from openwrt-packages is being built
<kyak>
i guess it means we can safely have our own cool version of ncurses :)
<kyak>
this means, build the wide-char ncurses only, cause it's binary compartible with one-byte ncurses
<xiangfu>
kyak: great.
<kristianpaul>
B_Lizzard: hey
<B_Lizzard>
Hey
<kristianpaul>
i noticed in my Ben that suspend on ram
<kristianpaul>
is not resuming after some hours
<kristianpaul>
the Ben just was power off
<zear>
maybe the battery runs out?
<kristianpaul>
nope
<kristianpaul>
i tought first
<kristianpaul>
but then turn it on
<kristianpaul>
and i was 68%
<kristianpaul>
s/i/it
<zear>
maybe the battery is loose
<zear>
try to shake your nn
<kristianpaul>
idid
<kristianpaul>
did witht the ben turned on
<kristianpaul>
and it dint poweroff
<zear>
i see
<kristianpaul>
apartently bat is good, but still some test to do
<kristianpaul>
i'm just saying what happen to me in, just in case some body else happened or even dind tried
<kristianpaul>
dint*
<kristianpaul>
btw i tried this suspend then let it there for few hours three times so far
<B_Lizzard>
Happens to me too
<B_Lizzard>
Probably a kernel problem, ask larsc
<kristianpaul>
larsc: hello
<kristianpaul>
There is a way of debug linux kernel in the Ben, using ie the serial console
<kristianpaul>
so i can save debug signals before the devices just power off...
<kristianpaul>
:-|
<larsc>
what do you want do debug?
<B_Lizzard>
The Nanonote shuts down after some time left in suspend
<B_Lizzard>
Probably crashes
<B_Lizzard>
2 - 3 hours
<B_Lizzard>
So if you do "echo mem > /sys/power/state" and leave the nanonote alone for some time it shuts down
<larsc>
as a software guy i would say it is a hardware problem ;)
<rafa>
larsc: :).. now you need to wait the answers from hardware guys
<wpwrak>
mirko: hmm, "new" image with Linux-2.6.32.16 ? :-(
<wpwrak>
rafa: which kernel version does jlime use ?
<rafa>
wpwrak: current beta versions built use 2.6.34.. version that you can built completely from OE (we some jlime devs already pushed stuff) uses 2.6.36 latest (with lars patches)
<rafa>
we some jlime devs= because some jlime devs
<wpwrak>
rafa: ah, so your upstream is more recent than your project's head. intersting ;-)
<wpwrak>
so one of those days i'll have to make my own 2.6.36 then ... see if MMC bind after unbind still hangs (like it does in 2.6.32.16)
<rafa>
cool no? :) .. but well, B_Lizzard and kristoffer are working to push the good ideas from beta and new ideas. And because lars did a big effort to put a lot of kernel stuff in 2.6.36 mainline why no to use it ? :)
<wpwrak>
oh, i'm all for tracking mainline closely :-)
<rafa>
wpwrak: working to push the good ideas from beta=Â Â working to push the good ideas from beta on OE mainline
<mirko>
wpwrak: what's the feature of 2.6.3X you miss?
<wpwrak>
mirko: dunno. i've noticed what looks like a bug but i don't know (yet) whether 2.6.36 is better or not
<wpwrak>
mirko: the bug is that, if i remove any uSD item, then  echo jz4740-mmc.0 >/sys/bus/platform/drivers/jz4740-mmc/unbind
<larsc>
wpwrak: just tested it and it works fine on 2.6.37-rc1
<wpwrak>
(with 2.6.32.16 (i think) per "official" september build)
<wpwrak>
larsc: ah, great, thanks !
<wpwrak>
time up upgrade those kernels then :)
<wpwrak>
(background: i've written a little library that should bring a bit more sanity into bit-banging - without introducing latency. one of the things it should do is guarantee that the mmc driver is moved out of the way. for this it used the unbind/bind.)
<wpwrak>
useS even
<mirko>
larsc: can we backport this fix?
<larsc>
well, there is no specific fix
<larsc>
you can backport the whole mmc driver
<larsc>
and see if it works
<wpwrak>
btw, has anyone on 2.6.32.x tried the unbind/bind, to check that it's indeed a kernel issue and not something weird i have locally ?
<xiangfu>
wpwrak: http://pastebin.ca/1987330 , tried several times, only once it's hang, but system not hang I can till open another ssh, and kill that thread.
<xiangfu>
wpwrak: by the way, what  is it use for??
<xiangfu>
time for sleep. see you tomorrow.
<wpwrak>
xiangfu: thanks ! yes, the system is still okay
<wpwrak>
xiangfu: what it does is that it tells the MMC driver in the kernel to release the MMC port (unbind)
<wpwrak>
xiangfu: or to re-acquire it (bind). this is necessary when using the uSD interface for bit-banging
<Ornotermes>
it's just a shame ST-Link don't work with linux
<wpwrak>
Ornotermes: admit it, you just want the bike ;-)
<kristianpaul>
ah ubind dint work with jliem kernel..
<kristianpaul>
thast why last night.. hmm
<wpwrak>
kristianpaul: you mean this ? echo jz4740-mmc.0 >/sys/bus/platform/drivers/jz4740-mmc/unbind
<wpwrak>
kristianpaul: or some other unbind ?
<kristianpaul>
thats yes
<kristianpaul>
welll not sureif that command exactly
<wpwrak>
hm, that ought to work
<kristianpaul>
i'll try again tnoight
<wpwrak>
i've never had unbind fail on me. the only thing that can go wrong is that you're trying to unbind after you've already unbound
<kristianpaul>
hmm
<wpwrak>
the kernel is nice enough to tell you, but none of the "echo" in openwrt feel this is information the user should be bothered with
<kristianpaul>
sure let me try again
<wpwrak>
not sure about jlime
<wpwrak>
(the echo problem is one of the subtle annoyances you get for using those "diet" environments, busybox et al. they often suppressed uncommon error paths, which can cause unexpected behaviour in advanced use)
<Ornotermes>
wpwrak: atleast i would have some use for a bike :P
<wpwrak>
(i don't know how much time i've wasted at openmoko with just this sort of problems. thus my dislike for busybox, dropbear, etc.)
<wpwrak>
Ornotermes: bah, for whose 2-3 days of sunny weather you get per year ;-)
<Ornotermes>
wpwrak: still better than a gas gusseling motor bike or half an apple computer :P
<wpwrak>
Ornotermes: though i have to admit that northern europe is impressive in summer. once been to oslo for a conference just in mid-summer. the city was full of convertibles and people greedily enjoying the few sunrays that graced it :)
<wpwrak>
Ornotermes: oh, i thought, with "bike", you meant the harley :)
<Ornotermes>
nah :P
<wpwrak>
Ornotermes: the apple bite indeed is an insult, i agree
<Ornotermes>
yeah, epecially since there is no stm32 ide for mac (as far as i know)
<wpwrak>
bah, the hip macs are for eloi ! what use would they have for morlock tools like an IDE ?
<Ornotermes>
eloi?
<wpwrak>
Ornotermes: H. G. Wells, "The Time Machine" ?
<Ornotermes>
never read
<wpwrak>
wolfgang: perhaps some idea for a t-shirt ? front side: <some fancy apple gadget> = Eloi, <something gritty from the qi-hw-universe>=Morlock. back side: dinner is served.
<wpwrak>
Ornotermes: shame on you. one of he GREAT classics. also available as movies. the old one is much better than the ~2000 version.
<wpwrak>
ther'es a lso a 1978 tv film, but i don't think i know that one
<Ornotermes>
actually i prefer movier rather than books, aleast they allow some multi tasking
<wpwrak>
you can drink beer while reading ...
<Ornotermes>
i don't drink bear at all :P
<wpwrak>
ah right, in sweden it would be vodka
<Ornotermes>
but if i would i could watch a movie, hack someting and drink a beer at the same time :P
<wpwrak>
GRRRR. the RF frontend I'm using with my USRP is only half-duplex, not full-duplex, as the documentation suggests. no surprise i only got weird results, no matter what i tried. naturally, all failure is silent ... :-(
<wpwrak>
(well, "documentation" may be a stretch. it's rather bits and pieces collected from all over the intertubes ... finally found the last bit in the mailing list archives)
<wpwrak>
kristianpaul: cute :) now, what happens if humidity changes ? ;-)
<wpwrak>
grmbl. crashy xchat
<rafa>
the safe version of jlime for qi/resellers does not have libmad .. and libsdl-mixer needs libmad.. :( .. so the most of sdl games will not work on that jlime version :(
<wpwrak>
kristianpaul: cute :) now, what happens if humidity changes ? ;-)
<kristianpaul>
wpwrak: well hope i dint changes :p
<wpwrak>
ah, the message made it :)
<kristianpaul>
rafa: WHAT?
<rafa>
kristianpaul: that.. what you read
<wpwrak>
"MPEG audio decoder library"
<kristianpaul>
damn patents..
<wpwrak>
rafa: does libsdl-mixer always use libmad ? or would it be possible to just replace the calls to libmad with abort() and still get something useful ?
<rafa>
kristianpaul: yeah :( .. but well, libsdl-mixer manages different formats.. one of them mp3 surely. WIthout libmad, libsdl-mixer is not installable.. so most of games will not work
<kristianpaul>
cause they use mp3?
<rafa>
wpwrak: we could avoid libmad into libsdl-mixer.. no all the games uses mp3
<kristianpaul>
just that?
<kristianpaul>
i meant is sound
<rafa>
wpwrak: libsdl-mixer manages different kind of sound formats
<wpwrak>
rafa: ah, zear mentioned that they're all quite mp3-happy
<rafa>
wpwrak: and many games do not use mp3.. so we could build libsdl-mixer without libmad support
<wpwrak>
rafa: maybe step 1: no sound. step 2: convert the mp3s to ogg ?
<rafa>
wpwrak: but well, I am not going to do this kind of stuff.. it is a little mess..
<kristianpaul>
ok so be nanonote =Â Â no targered for current games
<wpwrak>
rafa: or is there more involved than just conversion ?
<rafa>
wpwrak: I will put the stuff like it is.
<wpwrak>
rafa: okay, good enough for step 0 ;)
<kristianpaul>
well i like my dict and gmu :-)
<rafa>
wpwrak: I mean.. I do not know games using mp3.. surely there are many.. but no into OE. When zear talk about those.. he talks about DINGOO WORLD
<wpwrak>
rogue doesn't need mp3. the rest are inferior imitations or irrelevant  anyway ;-)
<rafa>
wpwrak: which is a COMPLETELY parallel linux-pirate world.. living at the same time that linux
<wpwrak>
ah ! ;-))
<rafa>
wpwrak: but most of games zear talks are not in any linux distribution
<kristianpaul>
whos saids pirate here??
<kristianpaul>
hides
<rafa>
they live into dingoo and similar pirate worlds
<rafa>
kristianpaul: I SAID
<wpwrak>
rafa: ah, i see. he had me scared for a bit with all the warez he wrote about
<kristianpaul>
said share instead
<kristianpaul>
:)
<kristianpaul>
say*
<kristianpaul>
any way..
<rafa>
wpwrak: they have fantastic games.. they share their codes between them.. most of them also do not know what "open source" or "free software" mean..
<rafa>
wpwrak: but well, into their worlds they share their codes.. that is nice at least
<wolfspraul>
rafa: sorry you lost me. So what is your idea for libmad & libsdl-mixer?
<kristianpaul>
btw any one can recomend a good lecture about logic analizers?
<wolfspraul>
can mp3 be configured out of libmad? or the libmad dependency be configured out of libsdl-mixer?
<wolfspraul>
or both will be left out?
<rafa>
wolfspraul: I will not do anything yet.. People trying to install games in jlime safe version will realize that libsdl-mixer will not install because libmad is missing.. just that so far.
<wpwrak>
wolfspraul: hmm, libmad is "MPEG audio decoder library" :)
<rafa>
missed*
<wolfspraul>
yes sure I know
<wolfspraul>
libmad is the #1 thing Sisvel is looking for on Linux
<rafa>
wolfspraul: so libmad is not part of jlime repo on qi/resellers repo
<wolfspraul>
and they enjoy the many dependencies to it... :-)
<wolfspraul>
rafa: can you create a list of packages that depends on libmad?
<wpwrak>
wolfspraul: so "configure mp3 out of libmad" doesn't sound like much of a choice. well, perhaps you could make a libmad-fake that does an open(`basename $1 .mp3`.ogg) :)
<wolfspraul>
wpwrak: yes exactly. although even the presence of this package name will make Sisvel look harder...
<kristianpaul>
rafa: what about games with no sound? is that posible? (even i know is not confortable)
<rafa>
kristianpaul: games without sound will work surely.. because they will not need libsdl-mixer installed.
<wolfspraul>
we need to be careful with naming. 'jlime safe version' and 'jlime full version' sounds like asking for trouble to me.
<wolfspraul>
so the 'safe' version is broken beyond usability, and everybody will first thing switch to the 'full' version, and the documentation for how to do that will pop up faster than we can delete it.
<wolfspraul>
risky.
<kristianpaul>
nolime
<wolfspraul>
maybe we find out the 'safe' version is not so 'safe' after all.
<kristianpaul>
sadly :'(
<rafa>
wolfspraul: yes, but well. IF you have some advices about names I am happy to know.. dont have enouhg english skills to be careful with names for example
<wolfspraul>
theoretically it doesn't matter. I'm just thinking ahead a little.
<wolfspraul>
if the 'safe' version is only to 'safely' pass customs, sooner or later we will fail.
<wolfspraul>
it also depends on how we look at this and how we use it.
<wolfspraul>
we need to be honest with ourselves :-)
<wpwrak>
"jlime for qi-hardware" ?
<wpwrak>
"official jlime ben edition"
<wpwrak>
etc. :)
<wolfspraul>
sure, anything goes
<kristianpaul>
offcial?
<kristianpaul>
just dont use the lime i think
<wpwrak>
(officially endorsed by the makers of the ben)
<wolfspraul>
I'm just saying we need to be careful to not assume that nobody will leave it installed anyway.
<kristianpaul>
sure
<wolfspraul>
because the 'upgrade' instructions will pop up everywhere
<kristianpaul>
already !
<kristianpaul>
just go jli.... com
<kristianpaul>
:p
<wpwrak>
wolfspraul: if we do this right, people may not even care all that much
<wpwrak>
wolfspraul: e.g., also document how to recode mp3 to ogg
<kristianpaul>
hmm
<wolfspraul>
correct they shouldn't. the 'qi/ben' image must be really usable.
<wpwrak>
there's only a fraction of the ben owners who wants to use it as an mp3 player anyway
<wolfspraul>
rafa: configuring libmad out of libsdl-mixer is too hard?
<kristianpaul>
wpwrak:Â Â only a fraction <- true
<wpwrak>
those who have mp3 via games don't really care about the format anyway. could be uuencoded high-res jpegs of 45 rpm discs ;-)
<wpwrak>
and also the number of people who really care all that much about games may not be all that huge
<mth>
are there games that use MP3? so far I've only seen WAV and Ogg being used
<wpwrak>
well, that's actually a question for the jlime folks: what's the "typical" user doing with their jlime-ben ?
<kristianpaul>
i wonder how many jlimes are running out thre..
<kristianpaul>
rafa: do you do stats about jliem donwloands?
<kristianpaul>
for example in beat4
<kristianpaul>
wpwrak: anyway ben is not a device for having bunch of sofware shiped i think
<wolfspraul>
kristianpaul: he, I think the whole 2 gb should be loaded with valuable free software and content :-)
<wpwrak>
kristianpaul: not YET ! ;-))
<wolfspraul>
if someone doesn't like it they can still reflash, but why ship empty bits - makes no sense to me
<rafa>
wolfspraul: set libsdl-mixer without libmad: maybe it is not hard.. if "configure" has some option to avoid libmad/mp* then it would be really easy
<wpwrak>
kristianpaul: i hope that will change with jlime :)
<rafa>
wolfspraul: two lists:
<kristianpaul>
wolfspraul: content is big :_9 for 2gb
<wpwrak>
wolfspraul: (ship empty) faster production process :)
<kristianpaul>
wpwrak: jlime change <-Â Â sure, seems go fast and realiable
<rafa>
kristianpaul: stats: no.. sorry. Maybe Kristoffer has those details.
<kristianpaul>
wolfspraul: btw the "digital color dictionary" uses the suspend on ram feature?
<kristianpaul>
or some sort of suspend?
<wolfspraul>
rafa: great lists, thanks!
<wolfspraul>
it seems libsdl-mixer is the only one that will take even more dependencies with it
<wolfspraul>
the others can just be removed I guess
<wolfspraul>
toppler has a direct dependency to libmad
<rafa>
python-mad could be used a lot in different packages perhaps.. let me check which packages need python-mad
<rafa>
no.. just 3 packages need python-mad
<rafa>
but well, the name of one of them is python-pymp3
<rafa>
which is not part of the repo I think
<rafa>
wolfspraul: about names.. I uploaded some image files with "safe" as part of the name :( Sorry, I will change that
<wpwrak>
(dependencies) doesn't look too horrible. if you don't tell anyone, they probably won't even notice ;-) well, duke and perhaps freeciv
<kristianpaul>
donwloading FEL
<rafa>
wolfspraul: wpwrak : libsdl-mixer built okey without mp* support (without libmad).. "configure" has some options to disable that. What would you advise? libsdl-mixer is needed just for few games.. You prefer to remove libsdl-mixer and the few games or you prefer try me uploading this libsdl-mixer version without libmad support and check what all work? (I mean.. somebody should test those games some day.. I guee that any of them use mp* .. so they would wor
<wolfspraul>
rafa: well it depends on what is maintainable for you
<wolfspraul>
I can only compare with OpenWrt. OpenWrt thankfully introduced a PATENTED config option upstream. So we have a baseline. Now if we need to drive configuration into some package, we can do so in a relatively clean and maintainable way.
<wolfspraul>
for OE, maybe you need to document the steps to remove patented technology somewhere, so it doesn't 'slip in' easily later
<rafa>
wolfspraul: well, it would be great to have into OE some config option like openwrt has for PATENTED. But maybe OE has and I do not know. But well, that is another regard I would say. For the current repo I would upload this libsdl-mixer version without libmad. I would like to do my best .. but if it gets messy soon (I do not know how broken is the repo with all the removed packages).
<rafa>
wolfspraul: unfortunately I will not have a lot more free time so I will not be so available to maintain something. I would like
<rafa>
to help to upload the current jlime version without patent problems.. but then.. users should use it like "it is"
<rafa>
because if nobody is going to work on that (jlime devs will work just on the jlime full version)
<rafa>
it will not get fixes from time to time.
<wolfspraul>
rafa: he, yes ;-) I can imagine and that's why I think about maintainability right away.
<rafa>
wolfspraul: yes, maybe the current repo is not a good idea on qi/resellers servers then. users willl find problems, they will complain, nobody will help.
<rafa>
current repo=curren jlime repo without packages with patented technologies.
<kristianpaul>
so stay as we are?
<rafa>
kristianpaul: I would say.. I am seeing that a broken repo is not a nice thing. Current jlime repo at jlime.com has some missed stuff.. which users find easily.. Now.. for the jlime repo at qi we would be removing several packages.. or many.. which does it still worst.
<rafa>
kristianpaul: Of course.. there are a lot of nice stuff on jlime repo which works.. for example all the development packages.. the whole perl/python.. all the X stuff.. several WMs... etc..
<rafa>
kristianpaul: but still.
<wpwrak>
rafa: but most problems people run into would exist in "full" jlime too. do you think the other jlime developers would refuse to help also with such cases ?
<wpwrak>
all that would be needed is a way to push or pull updated packages from "full" jlime
<rafa>
wpwrak: I think that they would help of course. BUt no for the special cases here.. The current repo has 16000+ packages.. so you have a huge amount of software ready to use.. But also.. some of them, or many (after we remove packages for the repo at qi) will not work surely.
<rafa>
And jlime people just will advise to use the full version in such cases I guess.
<wpwrak>
well, if they don't work, that would be a problem in the OE metadata anyway.
<wpwrak>
if you can't even install them, that's another story, of course
<rafa>
wpwrak: yes, I am talking about packages which you can not install because we removed packages. users find that easily :)
<wpwrak>
but if a package can't be installed, it should be straightforward to identify the reason, too. after all, we know which packages had to be removed.
<wpwrak>
i guess one should just walk the dependency graph and remove all packages that have a dependency that can't be resolved
<wpwrak>
don't you already sort of have that, with your opkg accelerator ?
<rafa>
wpwrak: okey, I will try with that idea before we show something about jlime as option for resellers/qi. No promises though
<wpwrak>
kewl, thanks !
<wpwrak>
what was the workflow so far ? something like this ? 1) manually identify all the problem packages. 2) remove the problem packages. 3) identify a few dependencies as well (and maybe remove them) ?
<wpwrak>
if yes, that should be relatively easy to automate. say, have an exclusion file that says  no libmad*, etc,. then kick out everything that depends on any package matching libmad* in some way
<wpwrak>
well, instead of kicking out, perhaps better to take the package list, remove the problems, and then output a list of packages that can be safely copied. rsync could then pick up those packages.
<wpwrak>
add a notification mechanism for any new additions (as opposed to mere updates) and the whole mirror could run almost on auto-pilot :)
<rafa>
wpwrak: it was : 1) I identified the problmatic packages. Got a list of them. Then 2) I wrote a simple script which removes the package and removes its entry from Packages index.
<wpwrak>
great. so only 3) is missing :-)
<wpwrak>
one small step for rafa, one great leap for qi-hw ;-)
<wolfspraul>
rafa: you said you will have less free time. got a new job?
<rafa>
wolfspraul: I ran out of money a time ago.. I am okey though.. we are two in this family ;) And I also get a bit of money of my previos employer who still has a debt with me and he pays me some coins from time to time. So I tried to find a proper linux dev job here in Bue, without luck. So I am evaluating options (I have worked as linux/unix admin for example, which is not nice in big companies, but that is something which exists)
<kristianpaul>
damn "warning: macro `AM_PATH_GLIB_2_0' not found in library"
<kristianpaul>
i think i broke my autotools some way :-/
<kristianpaul>
i cant run a autogen and not get erros..
<rafa>
kristianpaul: but that is a warning.. which else you get as output?
<wpwrak>
sound like the kind of warning that is usually followed by a segfault within a few millisecond ;-)
<wpwrak>
soundS
<kristianpaul>
i tought i may broke it when tring rtems first time..
<kristianpaul>
i was not carefull with configure install got some gnu stuff