<qi-bot>
[commit] Xiangfu Liu: [config.full_system] add gcc-mips, fixed in openwrt package: 8dbf61c, tested in build hosts http://qi-hw.com/p/openwrt-xburst/1c3892a
<kyak>
xiangfu: just for info, gcc-mips still build fine on x86 (in case you couldn't test it)
<xiangfu>
kyak: thanks.
<kyak>
no, thank you! :)
<xiangfu>
kyak: Hi I am thinking set the mac address and mtd partitions in kernel cmdline.
<kyak>
xiangfu: does it mean that mac address will be the same for all Bens? This doesn't sound very good.. Is there a way to make it unique for every Ben, but persistent for every device? (right now it changes after every reflash)
<kyak>
xiangfu: for mtd partitions in kernel cmdline - would do you mean exactly?
<kyak>
xiangfu: how does it help to specify the mtd partitions at kernel cmdline?
<xiangfu>
kyak: there are some people may don't want use root(512), data(1.5G). they want keep the old partition.
<xiangfu>
or someone want 2G for the whole rootfs.
<kyak>
so the partitions won't be hardcoded?
<kyak>
this is a very good idea.. just to make sure that reflash_ben.sh won't spoil data partition with rootfs data..
<xiangfu>
kyak: oh.. yes. correct.
<kyak>
maybe reflash_ben.sh can read the partition data from som file, which could be generated during build process
<xiangfu>
kyak: I think we just keep the reflash_ben.sh like that. but change name to reflash_openwrt.sh,
<xiangfu>
then people know this scripts file is just for reflash_openwrt. not for Debian or jlime.
<kyak>
current reflash_ben.sh (on the web site) will flash 256 Mb roots
<kyak>
but the one in git is modified for 512 Mb
<kyak>
how we can handle that?
<xiangfu>
kyak: oh. sorry. I forget to update, reflash_ben.sh in website is a link to [history] branch.
<xiangfu>
that's why the reflash_ben.sh is still 256Mb.
<kyak>
yes, and this is OK, because we don't have an image with 512 Mb rootfs yet
<kyak>
we just need to be consistent somehow with reflash_ben.sh... make sure that the correct version of reflash_ben.sh is used with 512Mb images (newer ones) and another one with 256 Mb
<kyak>
xiangfu: is there a way to modify kernel cmdline without booting into the system?
<kyak>
i think those who want 256 Mb or 2Gb rootfs will have to rebuild image anyway
<kyak>
so maybe giving mtd partitions via kernel cmdline is not needed after all?
<xiangfu>
kyak: no. they can use fw_setenv to set the kernel command line
<kyak>
but they have to boot into system for that?
<kyak>
and this booting may or may not be sucessfull depending on rootfs size
<xiangfu>
yes. 256Mb or 2Gb rootfs must rebuild the .ubi file
<xiangfu>
yes. they have to boot into system for modify kernel command line.
<xiangfu>
kyak: if he want change the u-boot or kernel partition, things got even worse .
<xiangfu>
kyak: thanks. for now just keep the partition hardcoded in kernel.
<kyak>
xiangfu: yup.. would be better to leave it like this
<kyak>
btw, i myself chechout the previous version of board-qi_lb60.c with 256 Mb :)
<kyak>
i wonder how much the boot time will increase with 512 Mb.. have you tried it?
<xiangfu>
kyak: not yet.
<kyak>
xiangfu: what about the mac address? is it hardcoded somewhere or what?
<xiangfu>
kyak: no. it's generate by kernel. I am not understand the kernel policy about the dynamical mac address.
<wpwrak>
xiangfu, kyak: you're about to enter the next level of complexity with NAND partitions :) if you choose to create the equivalent of a partition table, all your flash procedures then have to treat this as "sacred" data.
<wpwrak>
before you even think of a technical solution for this, perhaps it's best to first consider how you handle cases where things go wrong. e.g., when a ben has lost its partition table, how do you recover ?
<wpwrak>
there are two issues: a) there may still be good data on the Ben you want to make recover by recovering the partition table. b) even if you decide to start from scratch, your usage (e.g., what stuff you install) may require a certain partition setup or you'll run into problems later.
<wpwrak>
of course, one possibility would be to consider partition sizes to be now fixed for all times, and adapt to this accordingly (e.g., if the rootfs overflows again in may 2011, you'd introduce /home/opt/ or whatever)
<wpwrak>
other "sacred" data would be the MAC address. again, let's first consider how you can recover it. e.g., is there any non-eraseable information, machine-readable or not, on the ben you can use to identify it ?
<kyak>
i would think about such place in USB chip
<wpwrak>
which USB chip ?
<kyak>
(like in NIC cards)
<kyak>
the one that provides USB connectivity on Ben.. dont' know what is it
<kyak>
but as xiangfu mentioned, the mac adress is set by ethernet_gadget driver
<kyak>
maybe this driver can identify the USB chip in unique way and generate a mac address for it
<wpwrak>
there's no extra chip for this :) the USB device functionality is a block in the Jz4720
<wpwrak>
all the USB-related external components are passive. some beads/chokes, capacitors, that's it. all the rest happens in the CPU.
<wpwrak>
there doesn't seem to be any sort of chip on the device - other than the NAND itself - that can hold persistent data
<wpwrak>
one problem with the NAND is that most of the flashing procedures assume that it's okay to just erase everything
<wpwrak>
so if you want "sacred" data, you basically have three choices: 1) change all the instructions to avoid erasing that data. 2) change the tools such that they won't erase it. 3) find a mechanism that doesn't depend on the NAND content.
<xiangfu>
wpwrak: oh... thanks Werner, I will try to focus on other bugs. :). just leave the nand partitions and mac address as it is :)
<kyak>
wpwrak: very good explanation, thanks :)
<xiangfu>
leave for awhile, back later.
<kyak>
wpwrak: do you know if maybe Jz4720 has some unique ID that can be read from it?
<kyak>
because if it does... or if there is another way to identify each Ben in unique way, then it might be possible to have a fixed mac address at th end
<wpwrak>
kyak: unfortunately, i don't know of any unique id. (but maybe someone else knows something ?)
<wpwrak>
kyak: as far as I know, there's no human-readable ID either. sticker with serial number, or such.
<kyak>
oh, maybe the serial number can be read from it with software?
<wpwrak>
one possibility for a key would perhaps be the order number or such. but i'm not sure that's convenient to retrieve. also, people with more than one ben may find this confusing :)
<wpwrak>
wolfspraul: btw, rafa is looking for you. (he needs a location to host the "safe" jlime packages.) he should be up within a couple of hours or so
<wolfspraul>
sure
<wolfspraul>
I'm back online now, no worries
<wolfspraul>
where do those mp3 packages come from?
<wpwrak>
kyak: some MCUs have a serial number that's stored at some safe place inside. but as far as i know, the ingenic cpu's don't have such a thing.
<wpwrak>
wolfspraul: dunno. maybe a standard openwrt build.
<wpwrak>
wolfspraul: that stuff is also quite old. are these packages even still considered valid yet ?
<zear>
hmm.. does the safe distro share the repository with the unsafe one, or not?
<wolfspraul>
wpwrak: yes they are old, maybe overlooked. anyway thanks for the headsup, I just deleted those for now.
<wolfspraul>
maybe we need to delete/rebuild all of them...
<wpwrak>
zear: for jlime ? i think it would be best to have a completely separate mirror of the "safe" distro, on *.qi-hardware.com
<zear>
yeah, the thing is, a lot of my game ports will have mp3 music support
<wpwrak>
zear: can't you convert the music ?
<zear>
if you're willing to rewrite the music engine for each of them ;)
<wpwrak>
zear: oh. each has their own MP3 decoder ?
<zear>
no, but i'm not a programmer, i don't know how to replace mp3 music with ogg tremor one
<zear>
additionally, a lot of them are engine reimplementations
<zear>
i don't know if you can fill a lawsuit if someone has an engine reimplementation of your game in his repo
<wpwrak>
wolfspraul: downloads is a little chaotic anyway. perhaps this will be an opportunity to rearrange it a bit. e.g., with more than one distro, perhaps more specific names could be used.
<wpwrak>
zear: aren't the games already modified to some extent ?
<zear>
a couple of them are fine, because their base on gpl'ed engines, like duke 3d, quake or doom, but some other, like HOMM2, or Commander Keen are rewritten from scratch because the original engines were never released
<wpwrak>
zear: and if you're talking about MP3/other codec patent owners: yes, even if the code is someone else's, they can still get you if you distribute it
<zear>
no, i'm talking about making games that reuse someone's datafiles and game logic
<zear>
despite being coded from scratch
<zear>
but being coded in order to imitate the original engine
<zear>
some of them even base on recompiled and studied original binaries of the games ;)
<wpwrak>
zear: i somehow think that changing from MP3 to something else would be the least of the worries there ;-)
<zear>
as the rest is up to the game makers, and in most cases they are already nonexistent, or the ownership rights are in hands of someone else? :D
<wpwrak>
zear: reverse-engineered engines may have their own set of legal problems. even worse if you need some pirated level files/artwork/etc. in order to actually use them.
<zear>
anyway, if they read this chat, they could sure sue qi-hardware if they had that games in their repo :D
<wpwrak>
zear: i may be wrong, but i have the impression that jlime has some things that aren't quite clean in this regard
<zear>
wpwrak, the problem is, you cannot tell which ones were reverse-engineered and which were not
<zear>
some authors are nice enough to admit they decompiled the original binary
<zear>
but some of them don't mention anything, so you can't be sure if they just written it from scratch, or based it on decompiled enginre
<wpwrak>
zear: maybe ignore the engine for now and consider the levels/artwork. if that one is openly distributed by the original maker, you're probably safe
<zear>
*engine
<zear>
hah, that's a bigger problem then
<zear>
no, even the gpled engines, like duke3d or doom, still require original game data from you
<wpwrak>
zear: if there's an engine but you're supposed to obtain these things from somewhere else, you're deep in the grey zone ...
<zear>
i know of no games that got their engine released on gpl along with their datafiles
<zear>
maybe some smaller fishes, but not the big titles
<wpwrak>
zear: original game data per se is okay, as long as the company that owns it has released it under terms that permit this kind of use and redistribution. or at least if they have issues a statement to that extent. some have done this.
<zear>
this should be fine as long as you don't include that datafiles in the ipk, am i right?
<zear>
well, none of the engine reimplementations that were made without the original authors releasing their code, will suffer from this kind of issue
<zear>
i mean, all of them :D
<wpwrak>
zear: it's something i would consider at least potentially dangerous. particularly if it's unlikely that you could have the artwork without pirating it. e.g., if the original game device has some sort of copy protection, all "cd images" would basically have to be pirated (there are exceptions, but they're often on uncertain legal ground)
<wpwrak>
zear: again, let's look at the artwork first. i think artwork is more dangerous than the engine itself.
<zear>
well, all the engine reimplementation authors are kind enough to keep/reimplement the original copy protections mechanisms
<wpwrak>
;-)))
<zear>
like, they will still ask you for the words on page X of the manual :D
<wpwrak>
"and here is floppy_laser_hole() ..." ;-)
<zear>
although you could simply comment out such a line :D
<zear>
well, to be honest i don't care of the "safe" flavour of jlime
<zear>
won't use it
<zear>
i will release all my games for the free as in freedom flavour
<zear>
and it will be up to safe maintainters to select only the safe games
<zear>
though i just wanted to raise the problem
<wpwrak>
zear: you don't have to use it :) but it's still bad if there's material that can get others in trouble. also, it's not so nice if jlime or qi-hardware gets the image of being strongly associated with warez
<wpwrak>
zear: that may scare away many customers
<zear>
don't worry, i'll never include any illegal files in my game ports
<zear>
and if they require some 3rd party datafiles, it's all up to the user to get the original floppy/cd
<wpwrak>
also, if there are non-free games, you're not actually propagating freedom by distributing them despite that being illegal.
<zear>
well, the engine reimplementation is gpl
<zear>
and i'm redistributing them without the datafiles
<wpwrak>
it would be much better if you'd instead make something similar but better - and free, of course
<zear>
so yes, it's still freedom, despite being unplayable without the datafiles :D
<wpwrak>
yeah, i think the data files are the key. if you can get them "cleanly", then i don't think there's a problem
<zear>
i understand the problem qi-hardware has, but i find it to be exaggerated, and i still find that engine reimplementations on gpl == freedom :P
<zear>
you can get them cleanly on ebay
<zear>
unless second hand legal datafiles are counted as piracy
<wpwrak>
ah, that's an interesting issue :) i think there are court decisions that go either way. but if you can get them cleanly on ebay, that would imply that there's no copy protection on them.
<wpwrak>
this would make them relatively safe
<zear>
another thing is, nobody has floppy drives anymore
<zear>
so if you buy a legal floppy copy, are you allowed to use the datafiles from a different source?
<wpwrak>
duh. that's a tricky one.
<wpwrak>
maybe check if the company still exists. with things on floppies, there's a good chance that nobody is around anymore to trouble you ;-)
<zear>
true :D
<zear>
yet another thing is, that people are unaware of the GPL license terms, i have seen a lawsuit where someone after a police raid was accused of using illegal software with the damage to "GPL" company
<wpwrak>
hmm. that sounds interesting :-)
<wpwrak>
did they sentence him to be drawn and quartered by wild gnus ?
<zear>
it simply says that person used two programs which caused the undetermined damage for the GNU General Public License Free Software Foundation company
<zear>
i like how using programs causes damage for the license, not the original authors ;)
<wpwrak>
zear: someone got something very wrong there ;)
<zear>
the guy definitely lost the case as in whole, there are a lot of other illegal programs he used in this lawsuit
<wpwrak>
zear: perhaps it would help to classify those games/engines into three categories: 1) engine and artwork are free or with explicit permission to use.
<wpwrak>
2) artwork comes from some copy-protected media or is otherwise known to be used without permission. 3) all others.
<wpwrak>
1) should be perfectly safe. they're good to promote the capabilities of the device, without incurring any risks, or exposing others to risks.
<zear>
i'd add a 4) where it is known the game was created by reverse engineering the original
<wpwrak>
2) are things qi-hardware probably shouldn't touch. doing otherwise would be reckless (exposing qi-hardware) and irresponsible (exposing others who trust qi-hardware to provide them with "safe" material)
<wpwrak>
3 and 4) these are a grey area. i think it would help to better understand what do to here if we could have such a classification, perhaps with additional details. (e.g., what is the most common way for getting artwork, what is the status of the engine, etc.)
<zear>
wpwrak, for example, Commander Genius (Commander Keen remake) uses only level maps and sounds from the original
<wpwrak>
as far as i know, most of the modern engines are indeed open. e.g., the doom and quake engine, no ?
<zear>
though the gfx and music are all remade (athough the gfx still looks like the original, just refreshed, so it still isn't legal i think)
<zear>
doom still requires original datafiles
<zear>
though there is the "freedoom" version of game
<zear>
which is fully gpl
<zear>
as for quake, they were nice enough to release maps on gpl
<zear>
so all you need to do is to remake the textures and models
<zear>
to have a fully opensource game
<wpwrak>
for 3 and 4), one could also try to find the current owners of the material, and ask them
<wpwrak>
chances are that you'll get a friendly response for at least some of the old material
<zear>
or a cease and desist letter ;P
<mth>
most of the "only engine was GPL-ed" games have a shareware version though
<mth>
you can use the data files from that
<wpwrak>
mth: ah yes, shareware is good, too
<zear>
mth, to some extent
<zear>
some of them might forbid you from preinstalling it on commercial devices
<wpwrak>
zear: (cease and desist) well, better get it before you've done something that could cost you or others money :)
<wpwrak>
xiangfu: btw, is the rootfs size change in the kernel or does it come via u-boot ? (with mtdparts)
<wpwrak>
rafa: by the way, regarding the rootfs size change, i would suggest that all distros make that change, but that the partition layout is then frozen. there are still only relatively few bens around, and most of them aren't exactly "mission-critical", so this still isn't a horrible time to make such a change.
<wpwrak>
rafa: but the more time passes, the worse the effect of such a change. so this should probably be the last time, too. (for the ben)
<kyak>
i downloaded it and there are separate changes for jz4740 and jz4750 (very different)
<kyak>
i wonder if this is helpfull for jz4720
<rafa>
wpwrak: yes, but is the idea nice?.. WEll, suppose that we change to reflect that change.. But why just 256MB for rootfs?
<rafa>
wpwrak: I would not agree, We have 2GB of repo of software .. which umcompresed who know.. so there is a nice amount of software to test/install whatever.. 256MB for me, when nand is 2GB is no the best idea.
<kyak>
i haven't noticed any issues with MPlayer other than this squeky noise when rewinding
<mirko>
xiangfu: hey
<wpwrak>
rafa: if i understand this correctly, the change is to go from 256 MB to 512 MB.
<mirko>
xiangfu: experiencing issues as well
<rafa>
wpwrak: zear : we are talking of different things I think.
<wpwrak>
kyak: maybe put a bit of oil in the tape drive ? ;-)
<mirko>
was there a note about that change (256 -> 512) except in the commit log?
<rafa>
wpwrak: zear: the safe jlime distro just will use the safe repo, no unsafe
<wpwrak>
mirko: there was a brief mail by xiangfu, informing of the change
<mirko>
oh, so i missed it
<rafa>
wpwrak: zear : zear, your games ported were not packaged yet, so nobody can install them from opkg or package manager.
<rafa>
wpwrak: zear : zear, just fheroes
<wpwrak>
mirko: if xiangfu was the sneaky kind, then you'd think that would have been his intention ;-))
<rafa>
wpwrak: zear : and no.. the odd games with mp3 inside or strange data files will not go to the safe repo.
<wpwrak>
rafa: excellent, thanks !
<wpwrak>
was already having nightmares of tons of warez piling up in jlime :)
<rafa>
wpwrak: (about 256MB->512MB).. bah!.. every day there are new news about :D
<mirko>
wpwrak: hmm?
<mirko>
wpwrak: what do you mean exactly? :)
<wpwrak>
mirko: well, such partition changes are usually a big deal. so someone who anticipates that may try to create facts before anyone else has time to complain.
<mirko>
wpwrak: yes, i totally agree
<mirko>
that's why i'm a bit unhapy with the situation right now
<mirko>
*unhappy
<wpwrak>
mirko: now, i don't see xiangfu doing this sort of thing, but the evil half of my brain finds the implementation pleasing anyway ;-)
<mirko>
we were discussing this change
<rafa>
mirko: please, keep just one layout for ever :)
<mirko>
well, i'd love to...
<mirko>
*sigh*
<wpwrak>
mirko: yup. i saw the discussion. the implementation was just a bit ... sudden.
<mirko>
indeed
<wpwrak>
rafa, mirko: for a stable layout, the question would be: what do you do if you overflow rootfs the next time ? do the distros have a means to install things in different locations ? e.g., /opt/openoffice instead of /usr/*/openoffice ?
<wpwrak>
sorry, libreoffice, of course
<wpwrak>
a stable layout would avoid a lot of complexity, namely the introduction of some sort of partition table, and making all the processes and tools compatible with it. partition tables aren't a big deal on non-NAND media, because the rules are old and everyone, including all the tools, know them. with NAND, there's no standard, so you get plenty of chaos ...
<kyak>
we can't seriously talk about stability right now, when Ben is being developed so actively
<kyak>
we are changing uClibc version here and there
<rafa>
wpwrak: for Ben I do not like to have rootfs/data.. 2GB for one partition and we will not talk about this again ;)
<kyak>
the git layout is being changed too (i hope it won't anymore)
<kyak>
right now there are Bens overthere that are not binarry compatible between each other
<kyak>
rafa: the boot time will increase dramaticlly
<kyak>
also, there will bo no safe place between reflashes
<wpwrak>
yup. these are the problems.
<wpwrak>
otherwise, i'd like the one big FS as well :)
<rafa>
kyak: how dramatically? With jlime running X+WM+Desktop on nand it is 15sec I would say. And we have not done anything to improve the time yet.
<rafa>
kristianpaul: for safe place between reflased.. yes.. that is something to check.
<rafa>
kyak: for safe place between reflased.. yes.. that is something to check.
<rafa>
kristianpaul: sorry :)
<kyak>
rafa: dunnoo.. it takes i think around 10 seconds to mount that 1.7 Gb partition
<mirko>
ubi needs to scan for it's end-of-fs-marker
<mirko>
like most nand-filesystems
<rafa>
kyak: it took around 6-7 seconds to mount.. yes I just checked (we use the whole partition). But anyway.. a few seconds more and boot is done.. I would not change that layout just to get a few seconds more.. I mean.. We (jlime) would not be happy to change that from time to time. And you could say "hey, nobody tells you that you should change your layout".. but then we would not working to do a dual-booting feature.. because if jlime can not mount the open
<rafa>
we would not working=we would not be working
<wpwrak>
a hard-coded layout only works if it doesn't change often, that's for sure. so you have three choices: 1) don't hard-code it but invents some sort of partition table (with the "sacred" data it brings, etc.), 2) decide on one layout and keep it "forever", 3) avoid partitioning and deal with the consequences
<wpwrak>
now, the door to which hell shall it be ? (-:C
<wpwrak>
rafa: btw, wolfgang should be here. you can ask him about the mirror location
<mirko>
i just talked to him... he is out for dinner and will be back soon
<wpwrak>
mirko: thanks
<wolfspraul>
rafa: what can I help with?
<rafa>
wolfspraul: 1TB of space
<rafa>
no, just kidding :D
<rafa>
wolfspraul: we would like to host a jlime repo without packages with patented technologies. For the safe jlime
<rafa>
version for qi / resellers.
<rafa>
wolfspraul: The requirements are:
<rafa>
- the repo is around 1.8GB of size so far. And without packages with patented
<rafa>
  technologies it would be a bit smaller.
<rafa>
- opkg uses http access, so we should be able to get packages with wget and the
<rafa>
  repo would be ready to use I think.
<rafa>
- you should copy the repo either with a script (we will give you it) which use
<rafa>
  a list of packages with patented technologies, or you do the mirror and
<rafa>
  remove the problematic packages of it.
<wolfspraul>
maybe put it on downloads.qi-hardware.com ?
<wpwrak>
rafa: maybe easier if wolfgang just gives you an account on whatever machine will host these things, so you can put the scrips there yourself ?
<wolfspraul>
of course :-)
<wolfspraul>
rafa: email me your pubkey and I give you root access, then you can do whatever you want
<rafa>
wolfspraul: okey, I will
<wpwrak>
what should the location be like ?
<wolfspraul>
I hope just removing will work, some packages may need different configuration options to be patent-free
<rafa>
wolfspraul: btw thanks a lot to support the idea
<wpwrak>
say that the ben does not pass fcc limits ? :)
<rafa>
wolfspraul: well, so far the best idea is just to remove those problematic packages. If many packages have the chance to be built to be patent-free is a hard work to do so far
<wpwrak>
of course, the later report in 2009_12_09_sharism_ben/ says it does. maybe some file needs moving ?
<wpwrak>
rafa: your removal method also removes them from the index, right ?
<rafa>
wpwrak: yes
<wolfspraul>
wpwrak: we had several go-arounds on the FCC cert
<wpwrak>
wolfspraul: so it seems :) the organization in certification/ is a bit confusing. FCC_Test_Report_20090901.pdf in the top-level suggests that it is the most relevant file, yet it doesn't seem to be
<wpwrak>
wolfspraul: also, 2009_09_30_final/ apparently isn't final
<rafa>
wpwrak: but no removes the packages which need the problematic packages. For example, suppose that A is a safe package, A needs of B installed. B is unsafe package. My method removes B. A will be there, but we can not install/use it, because B is removed.
<wpwrak>
rafa: okay, looks like a feature for version 2.0 :-)
<wpwrak>
rafa: it would also be better for the user experience if you don't offer packages they can't install
<rafa>
wpwrak: yes, of course. Feature for 2.0 :)
<wpwrak>
wolfspraul: now, software/ ... should jlime live there as software/jlime/ ? if yes, would it be confusing to have software/images/ and software/packages/ that are not compatible with jlime, yet use a name that suggests they are somehow general ?
<wpwrak>
wolfspraul: no more confusion about false screen protectors ;-)
<wolfspraul>
wpwrak: I have no problem with renaming but we need to communicate it properly.
<wolfspraul>
we just had this case with the 256->512MB rootfs change that wasn't communicated well
<wolfspraul>
wpwrak: [picture] yeah, I like that! that is clear! :-)
<wpwrak>
wolfspraul: (communicate) sure. if there's a rename, maybe even have a transition period with a link
<drizzt>
hi all !
<drizzt>
much more populated than #gta02-core !
<wolfspraul>
drizzt: hi :-)
<wpwrak>
drizzt: indeed :)
<drizzt>
hi wpwrak :)
<wpwrak>
drizzt: joining the ranks of the happy owners of tiny linux devices with a keyboard ? :)
<drizzt>
nano notes ?
<wpwrak>
aye
<drizzt>
not yet, got too much to do, not even got time to play with my freerunner anymore
<wolfspraul>
drizzt: what? not a NanoNote owner?
<wolfspraul>
hmm
<wolfspraul>
our bot should take care of that...
<drizzt>
ok, email sent
<drizzt>
feel free to ask for information
<kyak>
according to Ingenic's site, WinCE is also ported to jz47
<kyak>
this is interesting
<zear>
kyak, well, since you can't get ahold of wince source code, trying to run such a port on the nn would be pointless
<zear>
while it would boot, there would be problems with keyboard, lcd and all the other drivers, imo
<kyak>
zear: ingenic provides something for this.. i'm not sure exactly what
<kyak>
some documentation at least :)
<drizzt>
wolfspraul:
<drizzt>
wpwrak:
<drizzt>
for the quotation request, could you give a first estimation (global) for tomorow ?
<drizzt>
we have a meeting with an "incubator" and he asked us for estimations
<drizzt>
what I have from other projects is between 30.000 and 60.000 euros
<drizzt>
(global estimations, for the development part only, from phytec)
<wolfspraul>
drizzt: for 30-60K EUR, you can expect us at least (!) to fly you to Mars and back
<drizzt>
lol
<wolfspraul>
Werner's new outer-space engine will make it possible... his current estimate is that it should be ready in 2 weeks. right wpwrak ?
<wpwrak>
drizzt: what would be the deliverables and the timeline ?
<wolfspraul>
wpwrak: he sent us a mail I think
<wolfspraul>
but I still need to log off now, long day...
<wolfspraul>
drizzt: don't worry you have an answer tomorrow
<wpwrak>
wolfspraul: actually, that's -2 weeks. with tachyons, you have to reverse the time vector.
<drizzt>
deliverables depend on what you do, but something like BOM, schematics and layout
<wpwrak>
drizzt: no prototypes, software, ... ?
<drizzt>
prototypes yes
<drizzt>
software it's my job
<wpwrak>
drizzt: you mention certification. i think you also need software for that.
<drizzt>
yep, I'm doing it for another project right now
<drizzt>
working on a TI davinci :)
<wpwrak>
drizzt: also, what would be the openness requirements ?
<drizzt>
for the timeline, 3 to six months would be great for first operationnal prototypes
<drizzt>
openness, if it depended on me alone, it would be copyleft
<wpwrak>
drizzt: okay, a) openness of input (components used, etc.), b) openness of output (design, etc.)
<wpwrak>
(prototypes) how many ?
<drizzt>
but I have two "partners", one working with free software already, but the other comming from the medical market, and it hard to make him remove the word"patents" from his vocabulary
<drizzt>
a) openness of input : preferably with no NDA for documentation
<wpwrak>
drizzt: do it like they do on tv - make a little device that emits a loud been when the foul word is uttered :)
<wpwrak>
beeP
<drizzt>
b) openness of output : actually a licence like CC-by-sa-nc
<wpwrak>
wolfspraul: i'm not familiar with any of the three socs. you ?
<drizzt>
prototypes ... at least three I think, maybe more depending on what tests we need to perform
<drizzt>
one more for each destructive test requiered, though I do not see any, but on the project I'm working for, they used some prototypes for destructive tests
<wolfspraul>
drizzt: I won't do anything -nc
<wpwrak>
yup, -nc is bad
<drizzt>
I'd like not to do -nc, but the problem is that we cannot spend 100 to 150Keuro on the project and see the device copied before we manage to sell it
<drizzt>
it's intended to a quite concurential market
<kyak>
wonders what is the mysterious device these guys are talking about
<drizzt>
but if you know of a way to keep the openness and prevent the bad case I just mentionned, I'll much more than happy :)
<drizzt>
I'll be much more than happy :)
<wpwrak>
it's usually surprisingly hard to get people to steal from you ;-)
<drizzt>
this alone would bring me to mars and back :)
<wpwrak>
besides, if someone is really after stealing, the -nc won't stop them
<drizzt>
kyak: let's say we'd like to do what openmoko did for the cellphone market, but on a much less mature market
<drizzt>
thus giving the project much more opportunities
<kyak>
what is this market?
<drizzt>
domotics
<kristianpaul>
win 17
<drizzt>
lol
<kristianpaul>
oops
<kyak>
sounds interesting
<drizzt>
domotics and tablet PC in fact
<wpwrak>
kyak: something like the arduino, but programmed in COBOL, for its close resemblance to natural language :)
<drizzt>
what we see in films, but actually costs an arm and an eye
<wpwrak>
kyak: 6-bit characters, of course. 0 is signed.
<kyak>
this market isn't free, but it definitely lacks some open-source solutions
<drizzt>
yes
<drizzt>
that's what brought our project up
<kyak>
wpwrak: you can make jokes and teach me something new at the same time, you know? :)
<wpwrak>
;-))
<drizzt>
so
<drizzt>
wpwrak:
<drizzt>
wolfspraul:
<drizzt>
let me know for the quotation, along with the conditions :)
<wolfspraul>
drizzt: yes
<wolfspraul>
thanks a lot for bringing it up!
<drizzt>
none are "out of order", I'll just have to work harder to make my partners see the interest
<wpwrak>
yup, pondering it ...
<wpwrak>
drizzt: for domotics, the system seem rather feature-rich. have you considered and rejected the idea of a simple controller (with appropriately high availability) combined with a pc for the more demanding stuff ?
<drizzt>
yes
<drizzt>
considered and rejected, it's already available
<drizzt>
(modules using USB to interface to a computer)
<wpwrak>
that alone would not speak against it :)
<wpwrak>
wouldn't have to be just a usb comm dongle. could be an independent controller with some sort of wireless
<wpwrak>
the independent controller would be the executive that's available all the time and allows basic interaction. anything fancier would go to the pc.
<drizzt>
yes, but the goal is to create something new and open, to drive the prices down and provide possibilities to people like who would like to have a cool device for domotics
<drizzt>
but not limited to domotics
<wpwrak>
but this would exactly drive prices down :-) one can assume the pc to exist already. so by simplifying the special-purpose box, you make the whole system cheaper
<drizzt>
(should be used for much more applications, let alone a tablet with real conectivity)
<wpwrak>
btw, do you also envision a line of power line modules and such, or would it interface with some existing (presumably proprietary) standard/architecture
<drizzt>
yes, but you do not want to have to turn on your computer to lower the shutters or change the temperature of one room
<wpwrak>
for this sort of operation, you'd probably want very low-cost local modules.
<drizzt>
it will be based on existing standards
<drizzt>
there's a few open ones, and some you can access just by buying the microcontroller to interface with
<wpwrak>
(i should mention that i've been contemplating a domotics solution for my apartment for a while. that's basically the architecture i have in my mind.)
<drizzt>
and I personnally don't care for the few others, should they become obsolete
<wpwrak>
most of the modules are quite expensive :-( well, X-10 is cheap, but then it's X-10 ...
<wpwrak>
use of existing modules saves you from all the certification trouble, which is a clear advantage
<drizzt>
yes, modules are expensive, that's part of why we'd like to build this device
<wpwrak>
ah, but even with this device, you still need the modules :) but yes, controllers are really expensive, too
<drizzt>
the modules will be integrated (that's the 3 to 5 microcontrollers), thus making the solution cheaper than each module sold separately
<wpwrak>
hmm, perhaps we're talking about different kinds of modules. what i mean are the things that go into the sockets, lamps, appliances, etc.
<drizzt>
and much cheaper than existing "partial" solutions wich are only one or two busses in a tablet, at something like 2 ou 3000 euros
<drizzt>
ho
<drizzt>
yes
<wpwrak>
they're distributed by nature. well, if you can recable everything you can concentrate them a bit more, but still ...
<drizzt>
I was speaking of the parts that controls them
<drizzt>
of course our solution does not remove the parts in each device you want/need to control
<drizzt>
but it will let you chose which module you want for each application
<wpwrak>
ah, you would connect to various bus systems. i see.
<drizzt>
actually, when you start using one "bus" (let's say one technology) you stick to one, and given the price, you cannot move to another
<drizzt>
yes
<drizzt>
that's the main point
<wpwrak>
okay, i see
<drizzt>
nice
<drizzt>
I need to go back to my work, but I'll stay connected, so ask any question, I'll be back in the evening
<drizzt>
++
<wpwrak>
why lithium for the battery ? wouldn't you want something that doesn't age ? i think lead is commonly used for this sort of standby battery in industrial settings (in case of blackout and such). not sure about what's common for domestic use, though.
<qbject>
wpwrak: Yes. 5000rpm spindle might not be appropriate for your areas of expertise, but it would suit me juuuust fine. :)
<qbject>
and there are definitely worse makes than Seig.
<wpwrak>
qbject: yeah, my environment limits me to smaller and less noisy machines. lest the villagers come visiting, carrying torches and pitchforks :)
<qbject>
s/Seig/Sieg
<qbject>
wpwrak: Werner, the electronic witch-doctor!
<wpwrak>
qbject: yeah, my little mill stays hidden. in its box, you don't even hear it much when i mill PCBs at 3 am ;-)
<rafa>
wpwrak: jlime zlib.. and IIRC qi openwrt as well
<wpwrak>
might be interesting to see if lzo could provide a similar speed boost. the size increase seems relatively benign.
<wpwrak>
of course, it all depends on cpu vs. nand speed. if nand speed is the bottleneck, lzo could even be slower than zlib, because you have to move more data.
<wpwrak>
rafa: btw, regarding the jlime mirror, did you get everything you need from wolfgang ?
<rafa>
wpwrak: I have not gotten anything yet :D
<rafa>
wpwrak: I sent him the pubkey as he asked me
<wpwrak>
rafa: bah, he's slow !
<wpwrak>
rafa: what else will you need ? i'm not sure anything has been decided on the location yet, or has it ?
<rafa>
No.. I need, after the mirror is done, to know the exact link of the repository, so I can set them properly into the image rootfs. Then I can build the tar.gz for SD and ubifs for NAND
<rafa>
Well, "No"= location has not decided yet I guess
<wpwrak>
ah yes, you need the link there too. right.
<wpwrak>
what location/URL would you suggest ?
<rafa>
wpwrak: I have no idea, because I do not know how flexible is qi web site or Wolfgang regard that :).. but.. I am okey with http://whatever../jlime/repository/Â Â .. for example, or another good one would be something like jlime.qi-..whatever../repository/
<wpwrak>
what else would live under jlime/, besides jlime/repository/ ?
<wpwrak>
i kinda like the look of jlime.qi-hardware.com. you could build all sorts of redirects based on this. not sure how flexible the setup is, though
<rafa>
wpwrak: (what else would live under jlime/): no idea, I just proposed that because it would give us a bit of flexibility if another thing needs to live some day ;) but well, as you say
<rafa>
jlime.qi-hardware.com would be cool as well if that is flexible
<wpwrak>
perhaps things like jlime/images/ ? or would they be under repository/ as well ?
<wpwrak>
seems that it's the mighty apache. so it's certainly flexible enough - if you know which of the HUGE_VAL^infinite+some_more options to tweak ....
<rafa>
wpwrak: yes.. /jlime/images/ would be a good thing to live under /jlime/. For example right now I have almost ready 4 at least. two beta4: one for nand, one for SD, and two bootstrap images: one for nand and one for SD. The bootstrap would be useful for people who does not want the whole X+WM+Desktop.. and just want to set his/her own environment installing packages from repo.
<wpwrak>
kewl. and you may also get things like jlime/testing/ etc. later on
<wpwrak>
rafa: hmm, have you ever used {character,string}{Color,RGBA} ? they don't seem to do anything :-(
<rafa>
wpwrak: sorry.. in which context?.. :) I do not know some API like that ..
<wpwrak>
SDL_gfx
<wpwrak>
actually, i should probably do my own text drawing functions anyway, because i need some really small font
<rafa>
wpwrak: for color I often try to use something like this : SDL_MapRGB(scr->format, 200, 200, 200)
<rafa>
wpwrak: scr is the SDL_Surface you get with SDL_SetVideoMode()
<rafa>
wpwrak: often=I think that I had problem with colors if I do not read the format of scr and you maprgb
<rafa>
wpwrak: or what is not working exactly?
<rafa>
wpwrak: do you want to put some text with specific color and size?
<wpwrak>
SDL_gfx has its own color format, RGBA. it works well for line drawing and polygons
<wpwrak>
just text doesn't seem to do anything
<wpwrak>
i'm trying to put text with the default font, any color (as long as it isn't black ;-)
<wpwrak>
well, i did try with a non-black background too, just in case :)
<rafa>
wpwrak: I will give you an example how I put text using SDL :
<wpwrak>
meanwhile, i've already hacked a quick and dirty 7-segment display. let's see how well this works ....
<qbject>
wpwrak: have you ever touch-typed on a keyboard with domes like the Ben?
<wpwrak>
qbject: hmm, on a simnilar one, the oqo
<qbject>
wpwrak: and that works with fingertips?
<wpwrak>
messily, but yes, it does :)
<qbject>
Messily?
<qbject>
How so?
<wpwrak>
nice, i invented a space alien font ;-)
<wpwrak>
drawing 7-segment letters with the middle and the bottom row swapped looks very interesting ;-)
<wpwrak>
qbject: (messily) you really have to concentrate
<qbject>
wpwrak: 1. Pictures of your new language? :)
<qbject>
wpwrak: 2. did you have to concentrate because of the oqo kb's small size, or because of the pressure required?
<wpwrak>
because of the small size
<qbject>
great. You got me thinking about my capacitive kb idea and I decided it wasn't that interesting, since it would surely violate patents even if I didn't use proprietary code to do it.
<qbject>
wpwrak : that's coming from a piece of equipment, right, not your Ben?
<wpwrak>
qbject: at the moment, yes. but the thing should also be able to run on the ben
<wpwrak>
(it's the received signal strength of a WPAN module)
<viric>
do you have anything working of that wpan?
<wpwrak>
i can send text strings around. (bare text in IEEE 802.15.4 packet, no IPv6 or such.) but the antennas aren't particularly good yet. i'm working on that at the moment.
<wpwrak>
(higher layers) there's a proper stack being worked on in linux-zigbee, so i don't really have to worry about these layers for now.
<wpwrak>
hmm, then i'd still need the MAC. my stuff is much more primitive. no addresses :)
<viric>
oh. :)
<wpwrak>
i do have the CRC, though. pure frivolous luxury, i know ...
<viric>
no forward error correction? :)
<wpwrak>
nope. that's not even part of the full stack. has a fast ACK/retransmit mechanism, though
<viric>
+ crc
<viric>
ok
<viric>
it may be interesting to see how many bits fail in a packet, though.
<wpwrak>
oh yes. in my tests, i only report CRC failure, but receive the sorry remains of the packet anyway
<viric>
ok
<wpwrak>
but they're usually beaten up pretty badly. the modulation seems to have a "sharp cliff" between good reception and junk. which is probably a good property.
<viric>
and makes sense to decide for a ack/repeat than any FEC.
<wpwrak>
yup, that too
<viric>
are wpan modules really cheap?
<wpwrak>
wanna know the BOM cost of atusd ? :)
<wpwrak>
viric: if yes, tell me for how many units
<wpwrak>
(note: the capacitors are probably not quite good enough. should probably use higher-grade items)
<viric>
so yes, that is what I would call 'cheap'
<wpwrak>
for 100 units, the price becomes USD 3.05 per unit (plus whatever the better caps cost)
<wpwrak>
what's missing in this calculation are PCB, case, manufacturing.
<viric>
clear
<wpwrak>
and distribution, taxes, etc. so maybe an end user price around USD 20-30 should be realistic.
<viric>
ah, that goes away from cheap :)
<viric>
but it's either paying that or spending time.
<viric>
Do you know if other devices with uSD could benefit from this?
<wpwrak>
atusb is more expensive. current single-unit cost USD 14.90. but that there are a number of things i need to rearrange. so it'll change. not sure in which direction, though :)
<wpwrak>
the uSD is just bit-bang, so it depends on whether the others have that capability.
<wpwrak>
the real communication that's happening with this board is a form of SPI.
<wpwrak>
but it's different from SD in SPI mode.
<viric>
SPI... I don't know what it means
<wpwrak>
SPI is a serial communication interface. signals MOSI (master->slave), MISO (slave->master), SCLK, nSS (slave select).