<wpwrak>
but i don't intend to spread patent knowledge. it's actually the contrary. i want clear and reasonably simple rules that you can apply without knowing all that stuff
<rafa>
wpwrak: yes, I think so.. and I have checked a time ago which packages have the offensive software inside.. so I can check my notes and to update those
<wolfspraul>
anybody can open a shop, collect some ip, start 'programs' to enforce/milk them etc.
<wolfspraul>
how do you want to explain this to anyone? especially rules to be safe from this, in general? impossible...
<wpwrak>
as a first try, you can just list the rules, without justification :)
<wpwrak>
or with very terse justification. some of these concepts can be considered public knowledge
<wolfspraul>
I think that's one reason why the FSF stays so strictly out of this, and only focus on copyright issues
<wolfspraul>
there might be sisvel-like companies in other industries too, like medical etc.
<wolfspraul>
I think if you want to be successful as a sisvel-like company, you do in fact need to focus on some industries, collect patent pools that are meaningfully and effectively enforceable etc.
<wolfspraul>
you cannot just randomly pick patents and industries, I think that wouldn't make money
<rafa>
I need some rest.. let me write the ideas in few hours.. so we can offer something from jlime ;)
<wpwrak>
you're probably right
<rafa>
wolfspraul: wpwrak : thanks a lot
<wpwrak>
but that doesn't matter for jlime :) what jlime needs is a set of action items that yield something you can accept. and what you can accept is largely your own judgement call, since there are so many gray areas
<wpwrak>
of course, we can analyze the situations, just like we did, so it doens't have to be completely arbitrary. but at the end of the day, it has to be what wolfgang, tuxbrain, probably pulster consider a tolerable risk level
<wolfspraul>
or we just accept the tax in some countries
<wolfspraul>
in fact, if Sisvel would make their licensing program even a little bit more effective, maybe people like pulster or tuxbrain would like to (!) choose that route
<wolfspraul>
so maybe we should encourage sisvel?
<wolfspraul>
make it easier for smaller shops to effectively pay
<wolfspraul>
online shop?
<wolfspraul>
accept credit cards?
<wolfspraul>
:-)
<wpwrak>
also, the link instructions are pretty clear, i think. a) don't link to jlime.com, except b) if it's something that is not related to the problem packages and isn't "above" material related to them in the hierarchy of the site. if you don't understand b), just follow a).
<wpwrak>
;-)
<wolfspraul>
it's a bit sad to punish 80% of the world for a strange system that the other 20% setup
<wolfspraul>
yes seriously!
<wolfspraul>
I am serious.
<wpwrak>
well, you mentioned the kind of lock-in they have in mind
<wolfspraul>
Pulster and Tuxbrain would happily pay the 3.50 USD / device rate, if they only could at those low volumes.
<wolfspraul>
the problem is a 6-figure down payment
<wpwrak>
their problem is that those patents expire soon. so they want to lock you into long-running contracts. once you sign these, you're trapped forever.
<wolfspraul>
and a contract that ties you forever into the sisvel world
<wpwrak>
exactly
<wpwrak>
even if they waive the down payment, they'd still have your balls
<wolfspraul>
sure, I am just describing another possible solution to the patent problem
<wolfspraul>
see it as a country-like tax
<wolfspraul>
we are not complaining about different VAT/import duties levels either
<wpwrak>
i still like my idea with the two islands and tall walls ;)
<wolfspraul>
or are we?
<wolfspraul>
so we are not
<wolfspraul>
why not let Sisvel tax certain things in certain countries?
<aisa>
wolfspraul: I came back to the computer specifically to post about my branch proposal.  Thank you for posting the link to IRC.
<aisa>
I'm off to bed, good night!
<wolfspraul>
I am telling you - _if_ joining the Sisvel licensing program would be easier, Openmoko, Pulster, Tuxbrain would all have joined long ago.
<wolfspraul>
we wouldn't even have this discussion
<wolfspraul>
gn8
<wpwrak>
yes, i suppose so. working around that mess also has a cost. and it's still risky.
<wolfspraul>
correct
<wolfspraul>
and we punish the 80%of the world that have nothing to do with this anyway
<wolfspraul>
but it's not an option, because Sisvel is not good enough :-)
<wolfspraul>
we should write an earnest letter to them to improve their system
<wolfspraul>
make it easier to join
<wolfspraul>
send small amounts effectively
<wpwrak>
of course, making a stand also sends a signal. no matter how small.
<wolfspraul>
the transaction costs at Sisvel are too high right now
<wolfspraul>
I want an online shop and credit card payments.
<wolfspraul>
then they ship me some key that I can use to sign certificates with, that will then let my shipment pass customs.
<wolfspraul>
come on Sisvel!
<wolfspraul>
invest in the future
<wolfspraul>
become a true global player
<wolfspraul>
private taxation models
<wpwrak>
in the 1970es and 1980es, there were lots of airplane hijackings. until it became standard procedure not to pay and to fight back instead. then that stopped fairly quickly. doing business with extortionists just makes thigs worse for everyone.
<wolfspraul>
bring it down to the level that even a shipment of 1 single unit can be effectively taxed
<wolfspraul>
seriously
<wolfspraul>
even governments would be interseted in this technology
<wolfspraul>
online shopping creates a big taxation problem in some countries
<wolfspraul>
but anyway, it's not an option because Sisvel is not good enough
<wolfspraul>
I guarantee you Tuxbran would solve the problem that way if it would be offered.
<wolfspraul>
but a 6-figure down payment? no
<wpwrak>
yup. luckily sisvel demand too much, so no way to yield to that temptation of making a pact with the devil :)
<wolfspraul>
:-)
<wolfspraul>
in a weird way, we are helping Sisvel extend their patents globally
<wolfspraul>
another approach would be to only do online shopping directly from Hong Kong or China
<wolfspraul>
they cannot catch such shipments
<wolfspraul>
I doubt dealextreme.com has ever heard of any 'patent problem' :-)
<wolfspraul>
I could then meet with the Sisvel guys from their HK office for some chit-chat. we would be neighbors.
<wolfspraul>
it's a strange world
<wpwrak>
;-)
<wolfspraul>
they would still laugh at me, because every day they make 10 times more money than I make in a month
<wolfspraul>
maybe they would feel sorry for me and offer a job as their sysadmin
<wpwrak>
so it would be only fair if they paid you a "thank you" for helping o extend their patents globally ;-)
<wolfspraul>
we do help them, really, because we take patents to seriously
<wolfspraul>
for them it's just a legal tool to make money
<wolfspraul>
we turn it into a big philosophical, legal, whatnot discussion
<wolfspraul>
all unpaid of course
<wolfspraul>
whereas they charge for every tiny move they make
<wolfspraul>
as our first mistake, we are elevating this shit to a global level
<wolfspraul>
which would broaden their market and income opportunities a lot
<wpwrak>
so what kind of asymmetric defense would you suggest ?
<wolfspraul>
don't know, just ranting...
<wolfspraul>
:-)
<wpwrak>
you could hold lectures about the evilness of patent companies in front of street thugs in HK. maybe some will get an inspiration ;-)
<wolfspraul>
you mean list some worthwhile kidnapping targets?
<wolfspraul>
kidnapping for ransom is an old profession in China
<wpwrak>
not a bad idea at all :)
<wolfspraul>
just need to have a lit of rich people and you can start with your business
<wolfspraul>
list
<wpwrak>
it would be in fact pretty much their business model, just in a more traditional way. ransom for ransom :)
<wolfspraul>
actually I do believe anti-patent lobbying is working
<wolfspraul>
in the end I think removing (temporarily, 20 years) and continuing with anti-patent lobbying is the right thing
<wolfspraul>
not sure though, I don't think about this all the time...
<wolfspraul>
let's see how the android & smartphone battles end up
<wpwrak>
the anti-patent lobbyists did score some wins, that's true
<wolfspraul>
in that field everybody seems to sue everybody else right now, I already lost track
<wolfspraul>
apple against htc
<wpwrak>
but the trolls may actually have more of an effect
<wolfspraul>
htc against apple
<wolfspraul>
microsoft against motorola
<wpwrak>
yeah :)
<wolfspraul>
oracle against google
<wolfspraul>
and so on and so on
<wolfspraul>
if you look at this and don't laugh, something must be wrong with your humor muscles
<wpwrak>
it's patently insane :)
<wolfspraul>
nokia against xyz too, right?
<wolfspraul>
I'm sure the legal department of the companies that have not yet entered the ring are a bit under pressure now.
<wolfspraul>
is RIM already suing someonw?
<wolfspraul>
maybe we should sue someone too, to look more professional?
<wpwrak>
nokia-apple
<wolfspraul>
I would have no problem printing "patents are evil" on each NanoNote
<wpwrak>
and apple-nokia. that was actually first, i think
<wolfspraul>
maybe we should maintain a league
<wolfspraul>
but over time those battles get messy, it might be hard to score accurately
<wpwrak>
(look more professional) you mean like at companies emphasizing all the cool patented technology they use in their products ? :)
<wolfspraul>
no, a bit more anarchic
<wolfspraul>
just sue
<wolfspraul>
press release
<wpwrak>
i think it's a group-think thing. while they're convinced that this is just the cost of a system that ultimately benefits them, they will keep at it
<wolfspraul>
"Qi Hardware is suing Apple, Microsoft, RIM"
<wpwrak>
even if it just drains them
<wolfspraul>
it looks so professional maybe some blogs will just carry it
<wpwrak>
do both: print the thing about evil patent then fine some ridiculous patent and sue them :)
<wpwrak>
fiLe
<wolfspraul>
nah, no filing
<wpwrak>
ah, fake press release. why not
<wolfspraul>
too much work
<wolfspraul>
just suing
<wolfspraul>
no fake press release
<wolfspraul>
its' real
<wolfspraul>
we sue
<wolfspraul>
we can write the claims on the wiki, collectively
<wolfspraul>
find out some addresses of courts, and off we go into the battle
<wpwrak>
don't you need to provide some material for filing the lawsuit ?
<wolfspraul>
:-)
<wolfspraul>
you take me too serious :-)
<wpwrak>
of course, when the thing is rejected to due formal errors, you can complain about *that*. shows how the small ones have no chance in the system :)
<wolfspraul>
we'll just be bystanders and watch with amazement what system some people have created to keep them fed
<wpwrak>
"we did our best to defend out IP but they didn't let us. instead they make ridiculous claims." (like that there's no patent :)
<wolfspraul>
yes!
<wolfspraul>
:-)
<wpwrak>
yeah. like a biologist dissecting parasitic worms.
<wolfspraul>
I think with Milkymist One and Xue we will learn about some new licensing companies.
<wpwrak>
"viel feind, viel ehr" ;-)
<wolfspraul>
wpwrak: did you do anything else on the schhist ghost hunting?
<wpwrak>
naw, haven't found where the ghosts are coming from yet
<wolfspraul>
ok it's on my list too, though not at the top
<wolfspraul>
we'll fix it eventually...
<wpwrak>
yeah. it's weird. but when juggling so many tools, weird things tend to happen.
<tuxbrain_away>
wow guys, so much patent literature to be early morning, well when we start suing :P, we can sue due  they don't sue us , and that's discrimination or monopolistic strategy or a secret sue club that in the large argues the big companies are sending encoding messages to pact prices and make detail reports of their parallel business with drugs, weapons and tomatoes
<wolfspraul>
tuxbrain: great idea
<wolfspraul>
we sue Google/Apple/Microsoft/etc for not suing us because it's a discrimination of our intellectual property
<tuxbrain>
hahaahah
<tuxbrain>
yeah that's brilliant
<tuxbrain>
good, I I had enough money to support this I will just do it for fun
<tuxbrain>
a pity we can see the faces of those guys recieving the sue....
<tuxbrain>
I have doen once, suing a big "house selling" company ,
<tuxbrain>
and wait in the door of the central to see their faces when the notary delivers the sue documentation
<tuxbrain>
It' was so funny
<tuxbrain>
expensive suit with ties olds mas come to that papers as a fishes when you given them food
<tuxbrain>
mas-> man
<tuxbrain>
in two hours they callme no renegotiate terms (long history .... but that moment worth 9 months of suffering)
<kristoffer>
morning
<tuxbrain>
morning kristoffer
<kristoffer>
hey
<tuxbrain>
if you review logs, this night (eastern time) wolfspraul and wpwrak, and lately rafa make a interesting discussion on jlime , NN, patents
<tuxbrain>
but I still with one doubt, about jlime final positioning about this
<tuxbrain>
As you also know I want to spread the wolrd how beautiful is jlime, in fact I also doing , next step is to directly sell NN with jlime by default in tuxbrain , but to do so and openly add links to jlime project every where etc , the damn patented issue must be solved, that the only thing stoping me of doing so (as you also know) last news was you where dissusing it but I don't anything far than that.
<kristoffer>
tuxbrain, In the discussion we had, we really didnt have much objections to removing patents. The deal-breaker was essentially that we understod that even if we removed everything, it wouldnt make any difference for qi in the sense of offering it as an optional installation/promotion on site. That would leave us in the position of giving up one of our benefits while not gaining much in terms of exposure. I a
<kristoffer>
m not fully clear on what areas the different distributors are handling (divided into europe/asia/america?). We are eager to get mainlined in the nn community, but it should also come with more promotional benefits.
<wpwrak>
tuxbrain: whether links to jlime.com are okay or not would be largely be a risk assessment on your side. wolfgang thinks they're dangerous. i would consider them just proper identification. (assuming the offensive stuff isn't right on top of jlime.com)
<wpwrak>
tuxbrain: but there's of course no right or wrong answer here. that is, unless you employ a telepath monitoring sisvel and friends
<wpwrak>
kristoffer: what we've been working out is just a mode of operation that would allow qi-hw to safely host jlime packages
<wpwrak>
tuxbrain: of course, another solution would be for jlime to not offer patent-encumbered packages or instructions for getting such stuff installed anyway on their site either. but as i understand rafa, that's not what jlime want to do.
<wpwrak>
wolfspraul: btw, it isn't necessarily that we're helping the parasites to usefully enforce their patents globally. it's more like acknowledging that the technology has been tainted, just like GIF, and then avoiding it
<kristoffer>
back
<kristoffer>
wpwrak, I read the logs of yesterday (rafa was kind enough to point them to me). I fully agree to your points made there.
<kristoffer>
wpwrak, for the benefit of nn (and everyone), getting choices out there for end-users is beneficial for everyone. If we need to comply (removed patents), we will do so. But it must be met with an  effort from qi by promotion and actually offering it as an option.
<wpwrak>
kristoffer: i think qi-hw would love to do this. but perhaps you should specify exactly what you expect in return, so that there are no misunderstandings.
<wpwrak>
(e.g., "promotion" is somewhat vague. also "offering" can mean many things, ranging from hosting the packages to selling devices with jlime preinstalled to exclusively preinstalling jlime)
<kristoffer>
wpwrak, agreed. Let me specify
<wpwrak>
(btw, just for the record, i don't speak for qi-hw, sharism, tuxbrain, etc. just trying to help to sort things out)
<kristoffer>
wpwrak, 1. Offering it as an option for sold nanonotes (e.g people can choose between openwrt/jlime/(debian?) at sale time(=offering of option), 2. mention of jlime on site as an valid option (=promotion).
<tuxbrain>
I'm not qi, due qi is a comunity , as tuxbrian I can tell than yes if you take that choise of remove patent, I will promote it, Tuxbrain is the EU distributor an AFAIK the second in number of sells after Sharism it self, and due if you are as savy as actual distro, there is no reason of put jlime content on the qi-wiki page or any other NN place
<tuxbrain>
and from my part choise will be put Openwrt, jlime will be by default on my NN :)
<wpwrak>
tuxbrain: (who is qi-hw) indeed, that's the next problem :) of course, the people operating the site could be considered as the "contract partners" there
<kristoffer>
wpwrak, tuxbrain I can understand that my approach sounds somewhat harsh or even arrogant. My main point is and has been, that users benefit from choice. And distributions benefit from competition.
<wpwrak>
tuxbrain: "no reason to put" or "no reason not to put" ? :)
<tuxbrain>
no reason to not put :) sorry
<wpwrak>
kristoffer: what would be "mention" ? just the name ? maybe name and logo ? would you expect a clickable link to jlime.com ? maybe some additional text or such ?
<wpwrak>
kristoffer: (1) i think wolfgang already wrote that he'd want to do this
<kristoffer>
wpwrak, Nothing much really. Just some text regarding there being many distributions for the nanonote and jlime being one of them. Im not looking for special treatment, just what all nn distributions should have.
<kristoffer>
wpwrak, I believe he said that in a perfect world he would want that, but its not practically possible for him to do so.
<tuxbrain>
I understand wolf position, he is a manufacturer his resources are limited an can no devote love to jlime, basically due he has no time/resources to not do so, and for coherence on his other products SIE/milky/xue, and whatever it will come for me is clever and I also support Openwrt as far as I can so the scarce resource it has on soft part are spend on Openwrt, but if jlime is as patent savy as oficial one, he is not again any one editing jlime p
<tuxbrain>
age on qi-hardware wiki
<wpwrak>
kristoffer: (your objectives) sounds very reasonable. the current situation that jlime officially doesn't exist and many people are uncertain if they can even mention it is clearly absurd
<qwebirc60090>
hi xiangfu
<qwebirc60090>
i tried paul link but no success
<kristoffer>
wpwrak, tuxbrain its a weird situation in any case. And finding common ground and clearing out the question marks is needed before anything is decided. The entire wiki question is another.
<wpwrak>
kristoffer: (perfect world) okay, we'd have to check with him then. i thought it was mainly a logistical issue, one that could be solved (e.g., they would have to stock devices with different preinstalled systems as different items at their fulfillment center, etc. so it would take some time to implement all this. my guess, with setting up the online order system, negotiating with the guys in HK, flashing and testing the devices, and s
<wpwrak>
hipping them there, probably something in the order of 1-2 months. that's of course just my personal guess.)
<wpwrak>
what would be "the wiki question" ?
<kristoffer>
wpwrak, if wolf would do that, then I would be quite happy.
<kristoffer>
wpwrak, well, in short. There was a mention that it was not a good thing to describe howto convert patent->open codecs on site
<kristoffer>
we support 4 different platforms (only one being mips), so we obviously want documentation for everyone.
<kristoffer>
If we comply to muffinman by removing patetnts and howtos, we then face the problem of having to remove them for other platforms as well.
<wpwrak>
that was about describing it on qi-hw. this means for qi-hw that qi-hw can't link to such descriptions
<kristoffer>
ah oki, then it shouldnt be a problem
<kristoffer>
wpwrak, lets hear what wolf says about this all once he returns (wakes up)
<wpwrak>
well, let me be specific. 1) qi-hw can't host such descriptions. 2) if someone puts such a description on the qihw wiki, mailing list, archive, or irc log, it would have to be removed from the archives. 3) if someone links to such information, the link would have to be removed from qi-hw. 4) wolfgang also considers it necessary not to link to jlime.com (or, let's generalize this, anywhere "above" the problem packages or mechanisms or i
<wpwrak>
nstructions for installing them)
<kristoffer>
wpwrak, so in short a link to jlime.com on site will never happen. Well, might not be a problem if the name atleast could be there, or a comment to search on google for it.
<wpwrak>
what i proposed was that qi-hw could simple host a package repository from jlime. that way, there's a clean separation. that repository could either be pushed from jlime or pulled from somewhere on the qi-hw side. (maybe by a cron job ;-)
<kristoffer>
wpwrak, thats not a bad idea. It would secure the option thing will still keeping it seperated (away from harm).
<kristoffer>
will/while
<wpwrak>
kristoffer: by the way, also the list of package repositories that comes with the "patent-free" distribution would have to include only the qi-hw repository. e.g., if people could install the "clean" distro but then simply  opkg install mp3-codecs , that would be bad
<wpwrak>
including a location with patented material, but commented out, would probably still be too close
<wpwrak>
on the other hand, if there's a nice article on jlime.com explaining how to switch to the jlime repository and then getting all the good stuff, that would be fine
<kristoffer>
wpwrak, agreed. I think its a good idea. If that could be merged with having muffinman as an option at buy time (or debian for that matter). I think everyone wouldbe happy.
<wpwrak>
excellent, thanks. let's see what wolfgang has to say about it then.
<qwebirc60090>
anybody worked on serial port communciation
<tuxbrain>
qwebirc60090:Â Â what's the question?
<qwebirc60090>
i need to have two way serial port comm between NN and PC
<kristoffer>
tuxbrain, also. Im sure we can work something out, never matter what wolf says later on.
<qwebirc60090>
it is having USB type A connector and loose wires at another end
<wpwrak>
wolfspraul: did you see the comments from kristoffer from a couple of hours ago ?
<qwebirc60090>
which end will be connected to NN and PC
<wolfspraul>
hmm, overloaded a bit
<wpwrak>
tuxbrain: heh, i was exaggerating a little :) but it should be relatively easy
<wolfspraul>
it is not true that Jlime couldn't be treated exactly like OpenWrt
<wolfspraul>
I said many times, if technology would permit, I would let my customers click on images in a drop-down selection when ordering a NanoNote
<tuxbrain>
yes, yes yes, try to sweet it now nerdy genius
<wolfspraul>
and pay a royalty to the people who uploaded whatever they selected
<wpwrak>
wolfspraul: yup. once the patent problem is gone, there's no reason for them not to be equals.
<wolfspraul>
this is the direction I am going to
<wolfspraul>
so since I cannot offer such a flexible upload/choose OS option yet, it's a bit more stop and go
<wolfspraul>
also granularity is bad
<wpwrak>
wolfspraul: (if technology would permit) kristoffer also wants that jlime would be offered as a preinstalled option, on par with openwrt. would that work ?
<wolfspraul>
I don't want to have multiple SKUs at the warehouse
<wolfspraul>
I don't want to travel to HK all the time reflashing back and forth
<wolfspraul>
of course, I am describing my long term vision first
<wolfspraul>
which is: the customer decides
<wolfspraul>
for practical reasons, I cannot travel to HK all the time, flash 100 units back and forth
<wolfspraul>
or even ever single one :-)
<wolfspraul>
how should this work...
<wolfspraul>
but we'll get there
<wpwrak>
i think it would also improve the out of box experience to have jlime. well, for those who choose it :)
<wolfspraul>
so for the time being, I am thinking about dual booting
<wpwrak>
it's always nice if the first contact is friendly. not getting hit over the head with having to set up communication, download a big distro, figure out how to reflash, etc.
<wpwrak>
ah, dual boot. okay.
<wolfspraul>
not because I prefer this or that software, or there is any 'official' one, but simply because of practical limitations of how I can reflash & sell, with the technology where we have it today
<wpwrak>
yes, that sounds interesting. could the two distros share the same /home ?
<wolfspraul>
the data handling is an open issue anyway (in OpenWrt)
<wolfspraul>
I am hoping/foreseeing a situation where we will ship quite a bit of (hopefully) valuable data on the device
<wpwrak>
is that more a "no" or more a "yes" ;-)
<wolfspraul>
such as dictionaries, offline wiki, maps, etc.
<wolfspraul>
right now we have a separate data partition, although I think almost nobody uses it
<wpwrak>
ah yes, that would be nice. also, better out of the box experience.
<wolfspraul>
yes sure it's moving in that direction
<wolfspraul>
but it's a lot of work and will need a lot of time...
<wpwrak>
so that's all ubifs, correct ?
<wolfspraul>
I do need an english-chinese dictionary on my nanonote, and I want to be able to easily install maps of cities I travel to, as needed
<wolfspraul>
(those are just my personal needs)
<wolfspraul>
yes, ubifs
<wpwrak>
rafa, kristoffer: does jlime support ubifs ? e.g., for sharing that data/home partition
<wolfspraul>
we could easily format that partition in jffs2 as well I would think
<wolfspraul>
it's unused right now anyway
<wpwrak>
from what i gather, kristoffer should be happy with dual boot. in fact, it's even better than being able to pick at order time, because it wouldn't force people to make a decision for which they may not have enough information at that time
<wpwrak>
i don't know much about ubifs. when i was following nand stuff, my impression was that it was very complex (at least compared to jffs2). how does mount time compare ?
<rafa>
wpwrak: yes
<rafa>
wpwrak: current muffinman version for SD does not have the ubi tools installed, you need ton install them
<wpwrak>
rafa: perfect. so dual boot would be relatively easy ?
<rafa>
wpwrak: what do you mean with dual boot?
<wolfspraul>
wpwrak: oh, ubifs is much better than jffs2 I think. mount time is much better too.
<wolfspraul>
openwrt boots in ca. 13 seconds
<wolfspraul>
but the rootfs partition is only 256 mb
<wolfspraul>
I believe last time I checked the boot time was 8 seconds longer with a 2gb rootfs partition
<wpwrak>
boot time is great, yes
<wolfspraul>
(those are all ubifs numbers, I have not used jffs2 in a long time)
<wpwrak>
rafa: a system that has openwrt and jlime and you can choose which to boot. you'd still want to share /home, for your own data and such.
<wolfspraul>
the only big question would be how to choose which to boot
<wolfspraul>
easiest is a menu, maybe in u-boot?
<wpwrak>
jffs2 does many horrible things. there are also these cleanup threads it spawns at boot/mount time that keep on running for a long time.
<wpwrak>
or just dump u-boot ? :)
<rafa>
wpwrak: ah.. okey.. yes.. all the tools to mount ubifs are on the repositories (just one package IIRC), but we will add that as default because several guys already asked how to mount and we forgot to add some doc in muffinman wiki
<wpwrak>
it's kinda ironic that qi doens't use qi or something similar
<rafa>
(about which package to install)
<wpwrak>
rafa: excellent. sorting all this out properly will certainly take some time, so there's no rush
<wolfspraul>
a dual-boot NanoNote would be cool I think :-)
<wolfspraul>
for me, Jlime would be for X, OpenWrt for console & framebuffer :-)
<rafa>
wpwrak: well, for ubi it is easy: install package, run two/three commands, and then you have ubifs mounted.
<wolfspraul>
then we add triple-booting including Iris later!
<wpwrak>
hmm, i wonder if the kernel image could conveniently contain a label for the boot loader to show
<wpwrak>
so that it doesn't need a priori knowledge of what things are installed and such.
<wpwrak>
too bad that there's no standard way of making partitions on NAND
<wpwrak>
let's see if CONFIG_LOCALVERSION makes it ...
<wpwrak>
... yes, it does. good. that would be a way to identify images.
<wpwrak>
(that is, uImages)
<wpwrak>
so if the kernel would be called, say, Linux-2.6.32.10-OpenWRT, the boot loader could just do a sed 's/\(.*\)-\(.*\)/\2 (\1)/'
<wpwrak>
idem for jlime
<wpwrak>
on real file systems, it could simply read a file saying what should be displayed
<wpwrak>
or use the same sort of convention for the kernel file name
<wpwrak>
bah. or just open the kernel file and read the version string. no need to be any different.
<rafa>
wpwrak: no following. .. what would bootloader do with that label on uImage?
<acathla>
Hi everyone
<wpwrak>
rafa: display it. so you know what you're booting. and the boot loader doesn't need configuration. or built-in knowledge of where which system lives.
<wpwrak>
instead, you'd just have a fixed number of kernel locations to look at, plus uSD
<rafa>
wpwrak: sorry, I do not understand. :(Â Â bootloader would display it when booting?.. which is the idea?.. after it is booting you will need to wait a ready system to do something.
<wpwrak>
you could even multi-thread :) so people wouldn't have to wait for the scan to complete.
<wolfspraul>
acathla: hey! there you are... :-)
<wpwrak>
rafa: for the boot menu
<acathla>
here I am wolfspraul  :)
<wolfspraul>
I have not followed fpgaarcade.com closely, if for the simple fact that his interesting blog has no rss/atom feed :-)
<wolfspraul>
but we've been mailing back and forth a few times
<rafa>
wpwrak: who would run a boot menu? .. the bootloader or some kern+initramfs+kexec?
<acathla>
I never used RSS...
<wolfspraul>
which fpga is on that board?
<acathla>
hum
<wpwrak>
rafa: that would have to be decided. there are many possibilities. a) make a simple boot loader (without menu) that either boots a "default choice" or a more extensive menu/recovery system, or b) make a more complex boot loader that implements a menu
<wpwrak>
rafa: in terms of political correctness b) would be preferrable
<wolfspraul>
adding dual-booting to u-boot seems like a relatively easy task, that I believe we can get to work within a few weeks or so
<wpwrak>
rafa: ah, a1) default is the "default" system, user choice is menu/recovery. a2) default is menu/recovery, user choice is "default" system. a2 would also be "correct". may be a bit inconvenient, though
<rafa>
wpwrak: I was thinking a simple idea for just two systems (one on nand, one on SD). If bootloader detects a SD then it would try to boot the system on SD, if there is not sd at that time, or it fails to find the kernel then it would boot the system on nand.
<rafa>
But, the only ugly thing is that from time to time you would like to boot from nand and if you forgot to remove sd maybe
<wpwrak>
wolfspraul: i would just drop u-boot. there's very little in it that has real value.
<rafa>
the bootloader boots that (if you have a system there). And things could be worst.. if both systems are identical you would not realize easily that you booted from sd and no from nand :)
<wolfspraul>
acathla: so for Milkymist One, are you interested in joining the project?
<wolfspraul>
what takes you here, what can we do for you?
<wpwrak>
wolfspraul: and on the ben, you're only using a tiny fraction of u-boot anyway
<acathla>
Spartan 3E 1200 wolfspraul
<wolfspraul>
ah OK
<acathla>
wolfspraul, i don't know if I can do something usefull yet
<wpwrak>
rafa: yes, making it put uSD before NAND would certainly be a possibility
<wolfspraul>
he
<wolfspraul>
acathla: if there's any type of people I truly admire, it's those that know that they know nothing
<acathla>
I have an AVnet FPGA card to start being comfortables with xilinx, but I only learned VHDL...
<wolfspraul>
like me
<wolfspraul>
:-)
<wolfspraul>
you have no idea how my head hurts with all this stuff we are moving into
<wolfspraul>
nice
<acathla>
i know i know nothing but I know I can learn :)
<wolfspraul>
we have a total of 3 FPGA related projects right now
<acathla>
but may be not learn Verilog...
<wolfspraul>
SIE, Milkymist One, Xue
<acathla>
SIE and Xue?
<wolfspraul>
I'll try to give you a quick status of all three
<wolfspraul>
'we' is complicated, a very diverse group of people around open & copyleft hardware
<wolfspraul>
South America, North America, Europe, Russia, China, Australia
<wolfspraul>
only Africa is under-represented
<wolfspraul>
that URL gives you some idea
<wolfspraul>
we just built ca. 50 SIE boards
<wolfspraul>
but they are sold out now, cannot make more right now
<wolfspraul>
work is happening towards SIE v3, maybe make a few hundred in a few months
<wolfspraul>
that board has a spartan-3 as well
<wolfspraul>
it's geared towards the academic/robotic world
<wolfspraul>
Milkymist One, we had a run of 6 boards, amazingly all 6 are functioning, and are currently being used for development
<wolfspraul>
we are now working on the second run of 35 units
<wolfspraul>
if you want you can get one, we are selling them for 350 USD
<wolfspraul>
we are trying to have them ready for the 27c3 congress in Berlin in late December
<wolfspraul>
Xue is a Milkymist derivative, using KiCad
<wolfspraul>
no Xue boards have ever been made, we would probably make an attempt in earnest after the next Milkymist One run
<wolfspraul>
that's the 3 boards :-)
<acathla>
ok
<wolfspraul>
Milkymist One (RC2) is probably the best and closest starting point if you are interested
<wpwrak>
wolfspraul: you;ll have to set up a live video feed from the xue run. first one with kicad :)
<wolfspraul>
bassel: you there?
<bassel>
hey wolfspraul
<bassel>
what up
<wolfspraul>
he, just wanted to see whether you are still alive
<wolfspraul>
or just your computer logging in automatically :-)
<bassel>
hehe :)
<wolfspraul>
did you fully recover now? I heard it got pretty bad...
<bassel>
yeah got back to life
<bassel>
yeah I'm fine now, but it went really bad
<wolfspraul>
acathla: when I mentioned the global Qi team, I forgot our front man in the middle east, bassel !
<wolfspraul>
our embargo breaker, he he
<wolfspraul>
(I should remove this from the log... :-))
<bassel>
:-)
<wolfspraul>
bassel: are you doing anything with the NanoNotes?
<wpwrak>
wolfspraul: make the bot that delete the previous line when it finds one saying "(I should remove this from the log)" ;-)
<acathla>
lot of people...
<wolfspraul>
for the SIE boards, unfortunately there were a few more hickups during the production run that one might have hoped, so I have no spare ones to offer you right now
<bassel>
yes I played with the board of one of them and a student of me in the hackerspace is trying to compile some memory app for the second one
<wolfspraul>
ah great!
<wolfspraul>
that's nice to hear
<bassel>
yes I was not active on the wiki yet
<wolfspraul>
for SIE, the maximum would have been 65, but we were only able to make 54 fully tested ones with reasonable efforts
<bassel>
really busy the last few weeks. but should do more on the wiki and mailing list side
<wolfspraul>
and those all went to the original buyers
<bassel>
ahhh that's bad to hear
<bassel>
when is the next round?
<wolfspraul>
relax things are moving here
<wolfspraul>
Carlos works on an improved V3 version
<wolfspraul>
the next round is when there is a customer who orders
<bassel>
there is a fpga developer joining our teams here :-)
<wolfspraul>
tuxbrain is telling me he might want to order several hundred in a few months
<wolfspraul>
bassel: yes, acathla :-)
<acathla>
wow
<bassel>
I should bring him here
<acathla>
i'm not yet a FPGA developper and I know only VHDL
<wolfspraul>
bassel: so bottom line I cannot get you a SIE right now
<acathla>
I saw milkmist is in verilog only...
<wolfspraul>
bassel: I think on the OpenWrt side, we will do more work to make it easier to generate customized images. That should help you.
<bassel>
wolfspraul it's ok
<wolfspraul>
also we continue to upstream stuff, it looks like finally xburst-tools is going into Debian, for example
<bassel>
I noticed that most people try to run debian on it. maybe will make things easier than openwrt
<bassel>
yeah that's the things
<wolfspraul>
it's a natural first reaction
<wolfspraul>
but I dont' think you will be happy with it, mostly because it's too slow
<wolfspraul>
a customized OpenWrt-based image is best for you, imho
<wolfspraul>
I'll do more work to make that easier, and to document the steps, and keep you posted
<acathla>
ok I go play with my small FPGA board, i have a lot of docs to process...
<wolfspraul>
acathla: stop by this channel once in a while if you like, we have a lot of friends and supporters of open/copyleft hardware here
<acathla>
wolfspraul, I also got a LCD TV with FPGAs on the board, I dream to hack it and be able to display pixels :) http://acathla.dyndns.org/toshiba/
<methril_work>
hi
<methril_work>
acathla, nice, if you hack your tv let me know
<kristianpaul>
acathla: hey
<kristianpaul>
ji
<kristianpaul>
hi
<methril_work>
i'm note enough encouraged to break a expensive toy
<tuxbrain>
methril focus... focus....
<wolfspraul>
acathla: yes, very interesting. 2 fpgas even. it shows how common fpgas are nowadays in consumer electronics.
<kristianpaul>
acathla: i learn soem vhdl too, know i'm learning verilog with milkymist :)
<wolfspraul>
and unlike before, they are not being replaced with asics anymore in many cases.
<kristianpaul>
acathla: why not start porting milkymsit SoC to your board? :)
<wolfspraul>
kristianpaul: he may even have the same avnet board you have?
<methril_work>
it's more powerfull and bug free work wit FPGAs
<tuxbrain>
I thinc acthla has an spartan 3, as SIE, not a 6 like milky, isn't it?
<methril_work>
if you did a FPGA bug, you could update with further low-level firmware
<wpwrak>
methril: or the contrary - you can ship it with more bugs, because you - at least hypothetically - can fix them in the field :)
<kristianpaul>
wolfspraul: is avnet yes, but i think and advanced model
<kristianpaul>
acathla: i have an Avnet Spartan-3A Evaluation Kit is nothing fancy but i can run a basic Milkymist SoC
<kristianpaul>
no milkydrop afaik
<methril_work>
wpwrak, well, it's also true :)
<kristianpaul>
also i'm not VJ guy :0_
<acathla>
wolfspraul, it's an old TV, i think now they give up and put ASICs, but to be sure I should void the warranty on my new TV and I won't do that :)
<wolfspraul>
acathla: how old is it? and how much did it cost when you bought it?
<acathla>
kristianpaul, it's the same card, it's not mine, I picked up one at the /tmp/lab, it' was used by lekernel to prototype milkymist I guess
<acathla>
wolfspraul, the 32" toshiba TV with FPGA is free, and not working. My TV is 1 year old now and not free at all :)
<kristianpaul>
any wayy. let me know how i can help you, as i'm learning too  both is better and fun :)
<methril_work>
kristianpaul, ii don't think the tesis is an introduction to verilog, i'll try to do some work on the avr and you need to hack a little bet before you start coding real things
<methril_work>
kristianpaul, oh! sorry, i though you told that lekernel's tesis was an introductory one
<methril_work>
i see the paper you put is an introduction to verilog :)
<kristianpaul>
methril_work: neither me
<kristianpaul>
methril_work: tesis is jsut to get an idea anot MM SoC as i said
<acathla>
kristianpaul, thank you but, later may be, i don't think I will learn verilog now, VHDL is enough to learn first. When I'll be a master in VHDL may be I'll try verilog, but is this really usefull? Can't we plug VHDL and verilog designs?
<kristianpaul>
acathla: not sure but in Milkymist no
<kristianpaul>
not technical issues just keep low bugs
<wpwrak>
kristoffer: so .. how does dual boot sound ?
<kristoffer>
wpwrak, dual boot is fair enough.
<wpwrak>
i think it's even better than an order-time choice. that way, people can compare directly
<rafa>
kristoffer: wpwrak : do you remember the iris idea?
<rafa>
IIRC the developer said that he was able to boot a linux kernel from it
<wpwrak>
rafa: yeah. i'm not sure what to think of iris, really. great hack value for sure, but is it sustainable ?
<rafa>
no idea.. I just was thinking on it as a bootloader with a menu ;)
<rafa>
to have dual boot
<wpwrak>
rafa: i'd rather look for something minimalistic. like the qi boot loader. tie it more closely to the kernel so that code can be shared. e.g., lcm initialization.
<wpwrak>
(tie closely) i.e., let it copy code out of the kernel. that way, the code only needs to be maintained at a single location.
<rafa>
qi boot loader: you mean qi bootloader from openmoko or the current uboot that qi is using?
<rafa>
as you can see.. I am really lost on bootloader land :)
<wpwrak>
the one from openmoko. that one was simple and clean. only things missing are lcm setup and the menu.
<wpwrak>
initializing the lcm would also mean that errors could be displayed. the lack of diagnostic output is the one main weakness of qi.
<wpwrak>
not sure if it would make sense to reuse the entire qi or to just cherry-pick. reusing the entire qi would mean to keep the arm and gta01/20 parts, which would add maintenance overhead
<wpwrak>
since they're of no direct use for qi-hw, a fork may be better.
<wpwrak>
also, by having a menu and interaction with the user, much of the partition selection logic could change
<wpwrak>
(reuse of kernel code) in fact, the code could be extracted on demand, with the extracted version living in the boot loader's repository. that way, one can build it without access to the kernel sources, but a maintainer could still easily synchronize.
<kristoffer>
wpwrak, even if u-boot is a binary mess. Its still more of a bootloader standard, so not sure if its a good idea to paint oneself in a corner by using a specific bootloader only for the ben.
<wpwrak>
well, the question is what u-boot really provides. on the ben, very few of its features are used. and its "scripting language" is not expressive enough to do many things with it, so you still end up having to add new command for special things.
<wpwrak>
so you'd basically end up with u-boot doing a few things you really need (like initialize the CPU and read NAND), then you have your system-specific code that's new anyway, and you have to interface with all the more generic than you really need interfaces inside u-boot.
<wpwrak>
so instead of trying to have a very complex boot loader that tries to be a small operating system, i think it's better to have a small and very specialized boot loader. just implement the bare essentials and keep the feature set capped.
<wpwrak>
e.g., consider lilo. it's still a perfectly viable choice and it had very little feature creep.
<wpwrak>
with u-boot, it may well happen that you first run the xburst initialization you wrote, then the NAND reader you wrote, and maybe also the menu you wrote. all this with only touching existing u-boot code where it glues your things together. so there's no real gain for you.
<wpwrak>
the only "good" big thing would be ext2. there's an ext2.c in qi and it's 940 lines. and that's actually lifted from u-boot. so you can just reuse that code and you already have everything useful u-boot provides.
<wpwrak>
a more subtle problem with u-boot is that you'll try to solve problems with the tools u-boot provides. this limits what you can do. e.g., when thinking of the issue of populating the boot menu out of the box, i immediately realized that one could use the kernel's ID string. that's something you can't do within current the world of u-boot. i had not thought of that before, when having to solve a similar problem, but that time with u-boo
<wpwrak>
t.
<osokuro>
Hi all.
<aisa>
hello osokuro
<kristianpaul>
hello
<osokuro>
I've been playing around with gmu. It's a very nice enhancement to my NN.
<osokuro>
Anyone in here using an NN for entertainment purposes?
<wpwrak>
osokuro: i suppose most of us do, in one way or another ;-)
<osokuro>
That's excellent! I'd seen mention of it on the list, but hadn't actually checked it out on the wiki.
<rafa>
I see from videos that it takes around 20 sec to boot on X.. maybe we should improve a bit that time so official image devs has some challenge to improve its 13sec boot time ;-)
<wpwrak>
rafa: i wonder if tuxbrain got all of the boxes "signed" this way
<wpwrak>
tuxbrain: btw, just curious, can you reveal how many nanowar editions you sold ? also, was there any pickup on regular ben sales, i.e., from people hearing about the launch, this way finding out about the ben, and choosing the regular model instead of the nanowar edition ?
<wpwrak>
tuxbrain: (background: i'm wondering if the nanowar launch actually reached people. think it may have been a great opportunity to increase visibility, but i'm not sure it really worked)
<osokuro>
I'm starting to feel really bad for all the OpenPandora hangers-on who looked down their noses at the NN. I've done at least half of the stuff I intended for OP on my NN instead, and that's just futzing around with it.
<wpwrak>
osokuro: i think nobody here would object to you having to feel even more sorry for them ;-)
<osokuro>
(Anybody else see on Twitter yesterday that the OP team's cargo van caught fire? It's comic how tragic the whole thing has become.)
<osokuro>
wb asia.
<wpwrak>
osokuro: there's definitely something to be said for wolfgang's approach of producing first and optimizing later
<osokuro>
wpwrak: agreed 100%. I was massively inspired by a similar previous effort called The 1980s.
<aisa>
thank you osokuro, I had to catch my train.
<osokuro>
gotta love trains. mine's still an hour off.
<osokuro>
wpwrak: i just hope we make it to an nn board with a solid usb host.
<osokuro>
production board, that is.
<wpwrak>
osokuro: i think that's a pretty certain feature for then ya
<wpwrak>
s/then/the/
<osokuro>
wpwrak: mm. is there any more enthusiastic timeframe for the ya than 'someday'?
<wpwrak>
i don't know of any. i think a few things have to happen before that. e.g., it would be good to have the ya in kicad as well. then, there's the question of case parts. can the existing ones be used and modified ? or do new ones have to be designed and made ?
<wpwrak>
then, wpan needs to see some real-life application, so that we can determine if it's okay to integrate
<wpwrak>
furthermore, there's a better cpu wolfgang wants to use. not sure about the status of that. afaik, it's still not clear when and in what packages/bonding variants it will be mass-produced.
<osokuro>
any idea if that other cpu will be compatible?
<wpwrak>
it'll be an xburst, but with two cores. not sure about the built-in peripherals. but in general, companies try to avoid unnecessarily radical changes there as well.
<osokuro>
Well that's promising.
<wpwrak>
so, almost certainly not binary compatible, but probably not an overly big deal
<osokuro>
ahh. oh well.
<osokuro>
speaking of cpus, i've been trying to better understand the relationship between the Loongson folks and MIPS. wolfgang said at one point that mips had "made them pay" but it looks more like both sides brought something to the table.
<osokuro>
it seems to me that mips benefits from the world knowing that loongson is mips-compatible.
<osokuro>
(and it had not been marketed like that before their negotiations)
<wpwrak>
i don't know what that deal was. i asked wolfgang and he doesn't know details either. so it's a bit of a mystery.
<wpwrak>
it could be anything, ranging from mips offering something attractive for a low price, just to maintain the impression, to mips indeed forcing them to pay up and just giving them some trinkets in return to save face
<wpwrak>
of course, both sides will agree to presenting the deal in a way that makes them both look like winners. so you can't quite trust those press releases.
<osokuro>
heh, i never do.
<osokuro>
I just want to build a nice tough notebook for writing&coding with a pixel qi screen and a qi-hardware mobo.
<wpwrak>
"a pixel qi screen" ? what's that ?
<osokuro>
pixel qi is the outfit that designed the screen on the olpc xo-1. it'll do backlit video or frontlit e-paper. apparently can be made in most lcd fabs (with their license, anyhow)
<osokuro>
I've got an olpc on extended loan from a friend and it's mighty cool.
<wpwrak>
aah, i see. yes, i've heard nice things about that display
<AlbertDS>
Am I being thick? I ran "df -h" on the console but I don't see the 2GB NAND. Do I need to format and mount it?
<aisa>
AlbertDS: The NAND is partitioned into 3 partitions.
<aisa>
a boot, root, and user space.
<aisa>
but by default, the user space is not formatted.
<osokuro>
there are great instructions on the wiki [rummage]