<kyak>
rafa: are you there? could please give the mencoder line to encode video files best suited for mplayer in Jlime? I know i can see it in README, but Ben is not at hand right now
<kyak>
rafa: nevermind, got it from irc logs (thanks tuxbrain2) :)
<kyak>
going to match some video in air plane, if it works well
<kyak>
watch even
<kyak>
btw, recentkly i left Ben running for a night. I can say this it can handle at least 12 hours of autonomous idling :)
<kyak>
(but i also have three additional batteries from my old Nokias just in case)
<aisa_>
The package I am trying to port requires pyexpat.
<aisa_>
Normally, this is part of the standard python distribution.
<aisa_>
I have found this module with make menuconfig,
<aisa_>
and I *think* after that I rebuild the python module:
<aisa_>
make package/feeds/packages/python-compile V=99
<aisa_>
is that the correct module to build pyexpat?
<xiangfu>
aisa_: by ./scritps/feeds search expat. it's give me :
<xiangfu>
python-expat            Python support for expat
<aisa_>
a new trick!  Hooray!
<aisa_>
Thank you xiangfu.
<xiangfu>
aisa_: so you better run this: make package/python-expat/compile V=99
<xiangfu>
aisa_: first you need select 'python-expat' in 'make menuconfig' before run 'make package....'
<aisa_>
Yes, good.  I've performed that step.
<aisa_>
I get the following error:
<aisa_>
make[1]: *** No rule to make target `package/python-expat/compile'.  Stop.
<larsc>
`make package/python/compile`
<aisa_>
dang, something happened to my build tree:
<aisa_>
/home/a/wa/tracking-xburst/staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.32/usr/bin/../lib/gcc/mipsel-openwrt-linux-uclibc/4.3.3/../../../../mipsel-openwrt-linux-uclibc/bin/ld: warning: ld-uClibc.so.0, needed by /home/a/wa/tracking-xburst/staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.32/usr/lib/libc.so, not found (try using -rpath or -rpath-link)
<aisa_>
This worked the last time I was here :-(
<aisa_>
trying to build libtool.
<aisa_>
I'll run make clean and start from the beginning unless someone knows what changed in the repo the last few days...
<xiangfu>
aisa_: never touch the toolchain section.
<derRichard>
hi
<aisa_>
xiangfu: Did I?
<aisa_>
hello derRichard!
<xiangfu>
aisa_: I mean no one have commit anything about libtool and toolchain
<xiangfu>
aisa_: I think there is something wrong with your 'make clean'. I never run 'make clean'
<aisa_>
I ran make clean, but it failed again, so I ran make distclean.
<aisa_>
and I'm trying one more time...
<aisa_>
xiangfu: On fidelio, python-expat compiled fine.  I'm not sure why my VM was trashed, but as it is rebuilding, and I have a package, I'll call it user error.
<derRichard>
does mplayer work on ben? ingenic seems to have a port
<aisa_>
derRichard: I haven't tried mplayer yet, so I'm not sure if it works.
<derRichard>
a videoplayer would be nice to have :)
<aisa_>
derRichard: What would you be using it for?  Would you mind talking about your use case?
<derRichard>
watching movies would be great on ben.
<aisa_>
You would in this case then have a movie on a microSD card, plug it in, and want to watch the movie off of it?
<derRichard>
yeah
<aisa_>
Got it.  Very interesting.  I wonder if the battery is up to a full length movie, streamed from the card and decompressed.
<aisa_>
In theory it should, but I haven't tested the battery at load yet.
<aisa_>
(I've had my NanoNote for <24 hours...)
<derRichard>
my nn plays musik up to 7 hours
<derRichard>
using gmp
<aisa_>
excellent.  You have the music on a microsd card?
<aisa_>
or is it on the on-board storage?
<derRichard>
for now only on the on-board storage :(
<derRichard>
i don't own a microsd card. because i don't have a real use case for it.
<derRichard>
btw: a really nice use case would be ben as a password database
<aisa_>
I was the one who replied with what I'm currently using, which was not satisfactory for you.
<aisa_>
I was trying to remember which password database I have tried,
<aisa_>
I seem to recall one that worked iteratively, such that it asked one question then the next, until it got an answer.
<aisa_>
I wonder if that one would have a smaller screen size requirement.
<derRichard>
ah ok. you are alan?
<aisa_>
I am.  My IRC nickname is from a long time ago.
<aisa_>
Today I would use my real name :-)
<derRichard>
my irc nick is also quite old. anyway. cmp (console password manager) has is also very strange to use. using crtl plus a lot other keys sucks on ben's keyboard.
<derRichard>
i can provide ipkgs for cmp. it was a pain in the ass to cross compile it
<aisa_>
I do plan on using a password manager on the NanoNote, but it is a fair bit down on my list yet.  I'm not ready to test anything right away.
<aisa_>
Let me follow that up by saying that I do think the NanoNote needs a password manager, preferably one that works well with the keyboard and screen.
<aisa_>
having anything is better than having nothing,
<aisa_>
and taking that and making it something easy to use is a necessary, but second, step.
<wpwrak>
derRichard: agreed on the password manager being an interesting application. that was about the first thing that crossed my mind when i saw the device.
<wpwrak>
derRichard: in a way, it turns one the ben's biggest shortcoming, lack of communication, into a feature :)
<derRichard>
that's true
<derRichard>
when i have some spare time i'll write my own passwort manager for ben using qt.
<wpwrak>
derRichard: however, in terms of practical use, i think the ability to "type" passwords automatically would be desirable. ideally, over a wireless interface (for ease of handling)
<wpwrak>
derRichard: hence my interest in wpan :)
<wpwrak>
derRichard: i'm not sure you want anything as complex as Qt in a password manager. ideally, a password manager would not even link to libc. unless, of course, you've reviewed it :)
<derRichard>
why?
<aisa_>
wpwrak: Are you going to wrap your own system calls?
<wpwrak>
aisa_: it's easy
<derRichard>
not linking against libc does not help against wrapping system calls...
<wpwrak>
derRichard: to avoid accidental exploits. e.g., when sensitive information gets stored where it is more persistent than one expects.
<aisa_>
derRichard: could you reword that double negative?  I can't parse it.
<wpwrak>
derRichard: i think aisa_Â Â means the little bits of code that jump into the system call handler
<aisa_>
wpwrak: just because you can doesn't mean you should, I'm not sure I understand why you want to avoid libC but then have to write bits of libC?
<derRichard>
anyway, when you cannot trust your os then you are fucked in any case...
<aisa_>
Would it not be better to validate/check those portions of libC and use them?
<derRichard>
and what about the kernel? the system calls are kernel stuff. libc helps you only to call them...
<wpwrak>
aisa_: well, there would be very little of libc you actually need. the simple things are easy to replace. the harder things can be a risk.
<bartbes>
aisa_: I think he means when you hook the more lower-level calls, i.e. the kernel ones
<wpwrak>
aisa_: they may be a lot bigger than strictly necessary for the password manager
<aisa_>
bartbes: Yes, I believe I understand.  I was saying that libC does this, and the argument that doing it yourself is more secure is not true on the face of it.
<bartbes>
anyway, yeah, if someone hooks your kernel, then you're fucked anyway
<bartbes>
the only way for it not to reveal your passwords then is to simply not hand them out
<aisa_>
wpwrak: this is true of any application that uses libC, so you have argued that nothing should use libC.
<derRichard>
quite all known linux worms/trojans/etc... are kernel based
<wpwrak>
derRichard: yes, you have to trust the kernel to do what it does right. and not entrust it things where it could do bad stuff. so, no unencrypted data every goes to the kernel.
<bartbes>
but the question arises, if you got that far already, why not force the program to reveal the passwords?
<wpwrak>
aisa_: most applications don't need that level of security
<wpwrak>
bartbes: you'd of course need signed binaries and establish a path of trust right from the boot process. i'm still trying to find out how to do this :)
<wpwrak>
bartbes: in a way, it would be a separate os, yes
<bartbes>
that way nothing could be stolen
<derRichard>
guys....please come back to the real world...
<aisa_>
I agree with derRichard.
<bartbes>
lol, I was joking
<bartbes>
I simply told him to make an application not doing anything
<wpwrak>
derRichard: libc isn't the biggest part of the work. the hard part are the gui and the crypto-protocols
<bartbes>
but.. why?
<bartbes>
seriously
<bartbes>
WHY?!
<wpwrak>
bartbes: why what ?
<bartbes>
why would you want to go through this much trouble
<derRichard>
wpwrak: what are you talking about? why should i care about libc anyway?
<wpwrak>
bartbes: because a password manager should be very very safe
<aisa_>
bartbes: I would ask what he was going to do with his passwords while he was going through this much trouble ;-)
<aisa_>
see, I assume a password manager should very, very, exist.
<derRichard>
using gpgme and qt you can create a nice and secure password maanger
<bartbes>
wpwrak: but the moment you enter it, it leaves the security of the application
<bartbes>
so all is for nothing
<wpwrak>
bartbes: the moment you enter one of the passwords in the manager into some other system, yes
<bartbes>
well, I doubt you store the password for the manager in the manager
<aisa_>
LOL.
<bartbes>
so you'd have to enter it somewhere else anyway
<wpwrak>
bartbes: in a way, you do of course :)
<aisa_>
in a way?  in what way don't you?
<bartbes>
oh.. of course.. the one time I do want to write a blogpost there is maintenance..
<wpwrak>
aisa_: i mean, you store information encrypted wit that password, plus information that allows you to validate what you encrypted
<wpwrak>
aisa_: assuming infinite computing resources, you can retrieve the password with this :)
<bartbes>
wpwrak: also, what would take the passwords anyway?
<wpwrak>
bartbes: someone stealing the device ? someone taking it, installing a keylogger ?
<bartbes>
you're scared of stuff so low-level I wonder if you can even get back up once you do get control there
<bartbes>
I guess the keylogger is the only thing you really have to worry about
<bartbes>
what else can you do with a stolen device
<wpwrak>
bartbes: you have to design security bottom-up. can't build a skyscraper on sand, sorry :)
<bartbes>
not what I meant
<bartbes>
I meant, if you ever dig a tunnel under it, how would you ever get back to the top floor with just your shovel?
<wpwrak>
bartbes: you also have to consider usability. it's relatively easy to design a system that keeps its secrets very well but that's also inconvenient to use.
<wpwrak>
bartbes: convenience requires compromises. e.g., a way to "unlock" the device such that it stays unlocked for a while
<bartbes>
sure
<bartbes>
but if you steal the device
<bartbes>
you either have direct access due to a 'flaw' like that
<bartbes>
or you have no access because you don't know how to unlock it
<bartbes>
(a password presumably)
<bartbes>
so, how do you obtain that password other than a keylogger?
<wpwrak>
bartbes: a stoled but poorly designed device may have copied of decrypted information in memory
<wpwrak>
copieS
<bartbes>
oh right, I remember that one test where they froze the ram
<bartbes>
but come on
<wpwrak>
dpm
<bartbes>
if you have liquid nitrogen you might as well torture the dude anyway
<wpwrak>
don't forget that a password manager is as valuable as the sum of all the passwords inside
<bartbes>
don't forget the weakest link will probably be the user
<wpwrak>
so you have to defend against the strongest attack that's likely
<bartbes>
except if you do happen to go for plaintext password storage
<bartbes>
:P
<derRichard>
i'm leaving now. this discussion makes no sense at all..
<wpwrak>
and yes, you better include various forms of extorsion in your plans :)
<bartbes>
wpwrak: as far as I can tell your biggest threat is a keylogger
<wpwrak>
bartbes: that one's indeed a big concern. hence the need for a trusted path with signed binaries.
<wpwrak>
hah, extorTion. now i'm starting with spanglish :)
<wpwrak>
bartbes: keyloggers could also be hardware-assisted. so there's also the question to what degree you could "harden" a ben against physical tampering.
<wpwrak>
bartbes: none of this matters much if all you want a little password safe for personal use that hardly anyone else uses, because then any attacker would have to research an attack specifically against your system, which is unlikely to happen
<wpwrak>
(any millionaires present on this channel should take the previous statement with a grain of salt, though :-)
<bartbes>
gtg
<wpwrak>
derRichard: maybe you'll perceive a greater need for tight security if you consider that you'd also want to be able to store things like ATM PINs, e-banking access codes, etc., in your password safe
<derRichard>
wpwrak: have you ever worked with cryptography in the *real* world?
<wpwrak>
derRichard: a bit, yes. i also keep up on developments in the area, on both sides.
<wpwrak>
derRichard: the crypto block per se is usually the least of your worries. you just take one that hasn't been degraded much yet. e.g., AES. what's already much harder is to use it properly. and then create an environment that doesn't leak.
<derRichard>
your arguments are ok. but total unrealistic. using ben as password storage makes sense when you use good crypto (eg. pgp using gpgme) and you make sure no one installs a keylogger
<wpwrak>
why are those things unrealistic ? imagine a ben sold as password safe. then it has to withstand quite a number of attacks. at least to the extent that you a) can still claim it is actually useful as a password safe, and b) that you can inform users about its limitations in a way that's understandable and useful to them.
<wpwrak>
also, protecting against keyloggers isn't trivial. in the absence of tamper-proofing, you could only be sure if the ben never leaves your sight for more than a minute or so.
<derRichard>
are you using a passwort store on your pc?
<derRichard>
*password
<wpwrak>
no, too unsafe
<derRichard>
okay. know it becomes clear...
<derRichard>
*now
<wpwrak>
what i do, though, is keep less critical items in lightly obfuscated files. hiding in plain view works relatively well :)
<wpwrak>
one problem with password stores and the like is that they become an easily identifiable target for attack
<wpwrak>
anyway, feel free to use a system with weak security :) as long as there aren't many people using it and you're not a particularly interesting target, you'll probably be fine
<derRichard>
this is bullshit. you have to learn that there is and will be never perfect security. deal with it
<wpwrak>
i think that's rather obvious :)
<wpwrak>
i also wonder why you're so upset. in what way do weak password saves benefit you ?
<derRichard>
wpwrak: i'm not upset. i'm busy. sorry
<wpwrak>
alright :)
<derRichard>
busy with using my weak systems ;-)
<bartbes>
wpwrak: also, I see you keep referring to "Security by obscurity"
<bartbes>
which is...
<bartbes>
weak to say the least
<bartbes>
(yes, dinner has successfully filled my stomach)
<wpwrak>
bartbes: security by obscurity works reasonably well in some cases, e.g., the "needle in the haystack" case. as long as you assess the attacker's ability to search that haystack correctly.
<bartbes>
well, you specifically worry about high-value targets
<wpwrak>
ah, but for these i do't recommend the haystack
<bartbes>
in which cases security by obscurity is almost an improbably way of securely storing data
<wpwrak>
i said i use the haystack approach for myself, because i'm not a high-value target
<bartbes>
because if we're talking about low-value targets you're overthinking anyway
<bartbes>
come on, if you're low value they probably won't even bother to try and hack your ben
<bartbes>
they'll be like "wtf, pmp? can't use it? DUMP"
<bartbes>
the only time people will hack is either because they want to hack, or you're high-value in which case security-by-obscurity is an invalid approach
<wpwrak>
derRichard: "secrets and lies" would be better. applied cryptography is about the algorithms themselves. (and yes, i've read both)
<wpwrak>
bartbes: wait ... what i'm after is a solution that can be generally applied. i don't know your bank account, but i want to be able to tell you that it's safe for you to store your e-banking password in that safe.
<bartbes>
I won't anyway
<bartbes>
:P
<bartbes>
but you are overthinking
<wpwrak>
bartbes: so such a password safe would contain all sorts of passwords, belonging to all sorts of people. and it would also become an interesting target for examination
<bartbes>
you are discussing ways to protect data that nobody will even try to touch unless they really want to hack you
<bartbes>
in which case they will succeed anyway
<wpwrak>
bartbes: just consider all the encrypting usb sticks that have been cracked. it's not because the people breaking them had valuable data they wanted to get at
<bartbes>
for the lulz I guess
<wpwrak>
mostly security researchers. i think c't magazine cracked one too, although i'm not entirely sure if they did it on their own.
<rafa>
kyak: yes, it has the ubi tools, but you need to install some package which i do not remember the name
<rafa>
maybe mfd tools or something like that
<bartbes>
also, do note I am not a security expert by any means
<wpwrak>
heh :-) i just renamed the file. then it worked. probably some uninitialized variable
<kristianpaul>
ohh
<kristianpaul>
rafa: there?
<wpwrak>
all the other nprog commands worked fine. also, after the .ubi failed, i repeated one of the others and that one still worked.
<kristianpaul>
wpwrak: btw have you tried the last usbboot?
<kristianpaul>
aka xburst-tools
<wpwrak>
hmm, i got it a few weeks ago. could be that something changed since then.
<kristianpaul>
well i tried compile the last one some weeks ago and asked me about what seems mipsel toolchain..
<wpwrak>
could be that this has been fixed. there has been a command-line parser change ~3 months ago.
<kristianpaul>
rafa: i want try wikipedia on jlime, there is a doc about how do it? i forgot how ... and i really need it now (this days) :D
<wpwrak>
ah well, it worked, that's all that matters :) now ... making that system behave nicely in an automated signal strength test ...
<kristianpaul>
ohh
<kristianpaul>
utomated signal strength test <- can you tell more about it?
<kristianpaul>
a
<wpwrak>
it's very simple. i make atspi-txrx send a bunch of messages, with the usrp2 receiving. all this then goes to a program that obtains the 90% percentile (to remove any noise around the peak)
<wpwrak>
the idea is to step the board a few times through all the channels and record how strong a signal the usrp sees.
<kristianpaul>
and how are you "debuggin" the signal in the nanonote side?
<kristianpaul>
ahh is from Ben to usrp..
<kristianpaul>
i see
<wpwrak>
i just send with atspi-txrx. i made a small extension ...
<qi-bot>
[commit] Werner Almesberger: atusd/ERRATA: started errata section for 100912. Nothing major so far. http://qi-hw.com/p/ben-wpan/b390b28
<qi-bot>
[commit] Werner Almesberger: atspi-txrx can now send repeatedly, cleans up on ^C, prints the power level. http://qi-hw.com/p/ben-wpan/c99ba2c
<wpwrak>
.. here it is :)
<wpwrak>
so it sends a message with defined tx power a few times. i just go through the usrp output and find the near-maximum. that should be at the carrier frequency.
<wpwrak>
i don't do an fft to pick the peak from the right area of the spectrum. gnuradio is still too obscure for me to tackle such a thing. (i pipe things through a pipe, which causes losses. so i can't to an fft on the data on the other end of the pipe. but just looking for peaks should do as well. there aren't any supposed to exist outside the carrier +/- 1 MHz area anyway)
<wpwrak>
in case i get any weird results, i can always check the spectrum manually
<kristianpaul>
pics !
<kristianpaul>
and data are wellcome just for curiosity
<kristianpaul>
adn for those dont have a usrp2 at home ;)
<wpwrak>
it's mainly scripts :)
<kristianpaul>
ahh
<kristianpaul>
i tought you said "check the spectrum "
<kristianpaul>
wpwrak: okay what do you expect from the results?
<wpwrak>
(check the spectrum) yup, with the usrp2 spectrum analyzer script. the one in the first picture
<kristianpaul>
you can do 2.43Ghz peak, good :) ?
<kristianpaul>
-10 db too weak??
<wpwrak>
oh, i don't know what the absolute values mean, if anything. this is an uncalibrated antenna.
<kristianpaul>
:/
<wpwrak>
what's important is that the difference between the maximum of the +/-1 MHz region is at least 20 dB above the maximum in the region that's more than 3.5 MHz away from the center
<wpwrak>
that's one of the requirements in IEEE 802.15.4
<wpwrak>
hmm. also in the current image, gmenu2x still messes with the clocks
<kristianpaul>
because framebuffer issues?
<wpwrak>
dunno. why it does it. just something tells me that, whatever the reason, it's not a good excuse for messing with all the system's clocks ...
<wpwrak>
particularly if the resulting clock isn't a multiple of 16 :-(
<kristianpaul>
hmm
<kristianpaul>
lets move to ncurses and problem solved :)
<wpwrak>
yeah :)
<kristianpaul>
you should try same stuff with Jlime just for curiosity and if you have time for
<wpwrak>
i'll just ask rafa when he's back :)
<kristianpaul>
he had a usrp2 too??
<kristianpaul>
wow
<wpwrak>
naw :) but he can tell me if they're doing something nasty with the clocks :)
<kristianpaul>
hahah good point !
<kristianpaul>
you think really linux could boot in 3 seconds and behave better in the nanonote?
<kristianpaul>
just wonder why cellphones run real time OS and not monilitc kernels..
<kristianpaul>
hmm may be android people is doing something to achieve whaty i said i terms of speed and performance
<wpwrak>
hmm ? i think the ben boots already quite quickly
<kristianpaul>
10 seconds..
<kristianpaul>
well yes :)
<kristianpaul>
i jsut noticed this slow as more i use it
<kristianpaul>
ignore my complains :)
<kristianpaul>
sure is it quick compared to other stuff like fre..
<wpwrak>
hmm, the data that crawls out of my measurements looks very weird. i'm quite curious if i get the same sort of results also on the second run.
<wpwrak>
naw. just very noise. seems that my filter doesn't perform too well.
<kristianpaul>
hmm
<wpwrak>
hmm. can't say i like the way those bursts look. they're drooping. let's see if the other boards do that too ...
<wpwrak>
ah, much nicer. seems that this board has a bug. well, maybe. the one with "nice" signals is almost 50% stronger than the one with ugly signals. grmbl.
<wpwrak>
err, the other way around
<wpwrak>
ironically, the two are supposed to be very similar. this will be fun ...
<kristianpaul>
hardware is fun, isnt? :)
<methril>
hi all
<kristianpaul>
hey methril
<wpwrak>
yeah, you never know what surprise will come next :)
<methril>
hardware is "hard"
<methril>
:)
<methril>
how are you doing?
<kristianpaul>
good dealing with verilog to get a signal generation tool
<kristianpaul>
i'm planning do what wpwrak but froma fpga board to the ben
<kristianpaul>
so more simple signals less pain o hope
<kristianpaul>
s/o/i
<kristianpaul>
methril: what about you?
<methril>
kristianpaul, so bussy with paid job, an other paid project coming soon
<methril>
i thought i was going to have some time to hack a little bit nanos or Milkymist,
<methril>
but i need to sleep less to get time to hack something :)
<methril>
so it's not bad
<methril>
i'm doing some FOSS development at work
<methril>
a port of a BDM dongle
<methril>
similar to a JTAG, but for Motorolla/Freescale uC
<methril>
it's another OpenHardware tool (USBDM) for that kind of architectures
<kristianpaul>
hi qwebirc89194
<qwebirc89194>
hey
<qwebirc89194>
gota set my nick
<qwebirc89194>
hold on
<elricsfate>
there we go
<elricsfate>
kristianpaul: hi paul
<kristianpaul>
hello
<elricsfate>
kristianpaul: just got my nanonote in the male. updating it now
<kristianpaul>
Good news  !
<elricsfate>
kristianpaul: absolutely
<kristianpaul>
yay
<kristianpaul>
xiangfu: thre is fuse support in openwrt xburst?
<kristianpaul>
i was about try sshfs with my Ben then i got peer lost conection not sure if is because dropbear instea of openssh
<xiangfu>
kristianpaul: here is the search result './scripts/feeds search fues':
<xiangfu>
confuse                  libConfuse is a configuration file parser library
<xiangfu>
curlftpfs                CurlFtpFS
<xiangfu>
owfs                    OWFS - fuse file system
<xiangfu>
psmisc                  proc utilities
<xiangfu>
sshfs                    SSHFS
<xiangfu>
I am not sure which one can make fuse work in NanoNote.
<aisa_>
I'm certainly willing to just solve this problem the best way I can see,
<aisa_>
but if you have some comment I'd like to hear it first.
<aisa_>
xiangfu: I see your reply now on gforth, and I see that David asked for it too.  Awesome.
<xiangfu>
aisa_: I saw your email about /var/...  (sorry, will back in 10 minutes)
<xiangfu>
aisa_: the /lib/preinit/10_essential_fs:19. this file is for mount /tmp folder. but I think modify this file is not good idea.
<aisa_>
I agree.
<xiangfu>
aisa_: is that file must in /var/lib/Â Â ?? how aobut /usr/share/makfa/...
<aisa_>
In a normal unix system, one would not put it in /usr/share/makfa,
<aisa_>
because the makfa program itself modifies the file,
<aisa_>
and /usr/share is supposed to be read-only.
<Textmode>
is apparently invisible.
<aisa_>
In this case, it is my best idea for a solution.
<aisa_>
so it is what I am going to use unless we have a better idea.
<aisa_>
in a perfect world,
<aisa_>
I would like some kind of /var overlay,
<aisa_>
such that the contents are copied from the /overlay/var to /var at boot.
<aisa_>
and I can place "permanent" /var files in the overlap.
<aisa_>
I'm not really ready to do that kind of architecture change yet.
<aisa_>
but I think I will make a note for this kind of work in the future.
<xiangfu>
aisa_: how about ~/.makfa/
<aisa_>
It turns out this is essentially how the program used to behave.
<aisa_>
I have some notion that I will create a non-root user to work on the machine,
<aisa_>
so I was not thinking of this as an option.
<xiangfu>
aisa_: ok. understand.
<aisa_>
but currently, as there is the root user, I can see that it works.
<aisa_>
I like the /usr/share/makfa option better than this, because of this reason.
<aisa_>
Do you have a preference?
<xiangfu>
aisa_: let me search "openwrt /var/ /temp issue" first. :)
<aisa_>
perfect.
<kristianpaul>
ohh ./feeds/packages/fftw3
<xiangfu>
aisa_: it's hard code to package/base-files/Makefile:378
<xiangfu>
aisa_: so I would also like the '/usr/share/makfa'.  if we need more /var/... in future. we can think about patch the upstream :)
<aisa_>
agreed.  I'll add the upstream /var/... patches to my backlog, but put them after getting a test environment running for the nanonote.
<aisa_>
and in the mean time, I'll update the makfa package to use /usr/share/makfa.
<aisa_>
Thank you!
<xiangfu>
aisa_: you are welcome :)
<xiangfu>
kristianpaul: hi. I test sshfs in old openwrt release. it not working. sorry I remember wrong.
<kristianpaul>
xiangfu: no problem
<kristianpaul>
xiangfu: i wonder if linux can emulate by USB a massive storage device??
<kristianpaul>
that will be awesome
<xiangfu>
kristianpaul: there is one option in kernel menuconfig :Â Â File-backed Storage Gadget
<xiangfu>
kristianpaul: there are many . I just found there is one option: "CDC Composite Device (Ethernet and ACM)"
<kristianpaul>
xiangfu: wow great
<kristianpaul>
File-backed Storage ... implements the USB Mass Storage class (as CB, CBI, or BBB), appearing to the host as a SCSI disk drive. This uses a file or block device as a backing store for the drive, like the "loop" driver. (Read about how to set up the backing store.) The 2.6 version also lets you change the backing file associated with a given LUN
<kristianpaul>
:D
<kristianpaul>
For example, you might specify the block device for an MMC card when that card is inserted (maybe using a hotplug script, or a GUI). Note that for MS-Windows interoperability, your backing store needs to be formatted with a DOS partition label. Data partitions can then be formatted with VFAT (for the widest interoperability) or some other filesystem (such as ext3 or reiserfs, for use with Linux