Log from #neuros IRC channel

keep it as FAT. it's likely our fault, since we should really account for that

can you file a bug about that

<nick_fn>

ok, I'll just upgrade later :-)

<nick_fn>

ok

[03:26]

nick_fn: thank you, we'll get fixed. probably it was tested only with larger CFs

[03:26]

<nick_fn>

I don't really understand bugzilla much - bugzilla.neurostechnology.com doesn't reference patches/svn commits, or anything like that?

[03:26]

and that check slipped by

[03:28]

nick_fn: no, we don't have such a system in place like it would be sane to do. we just reference the date of the commit usually, since testing team works on release of the day before

[03:28]

<nick_fn>

Why not have bugzilla record the svn changeset numbers? (just curious, really)

[03:29]

automatically ? i have no idea how to do that

[03:30]

<nick_fn>

automatically is better (we do that), but manually is great compared to someone saying "bug is fixed" but without a link to the actual change

[03:30]

how do you do that automatically ?

<nick_fn>

svn commit hook

and in the commit msg you need to name the bug in some specific way ?

<nick_fn>

yes

<nick_fn>

It's a very popular way. Debian does it, we do it :-)

<nick_fn>

Much less work for the developers, and you get a great historical resource.

have a link to scripts used to do it handy ?

i agree it's useful

having team use it properly is another matter entirely though

<nick_fn>

only for Trac these days - but google 'scmbug'

will do

<nick_fn>

so now, I just say "Change the widget to blue; this fixes #230"

<nick_fn>

which automatically updates ticket 230, and changes it into the next stage of the bug lifecycle

that would be bad

<nick_fn>

why?

we use a bit of a non-standard bugzilla lifecycle

bugs get reassigned to lead tester when fixed, and he close them or reassing back

<nick_fn>

well, it's up to you to make things work for you - you'd just edit the integration script.

<nick_fn>

that's fine.

yes, i mean, bad just the part to move to next stage

the "linking bug to patch" is ok

where does it put the link ? in a comment ?

[03:35]

<nick_fn>

we have the commit messages be attached to the tickets as a comment, yep

[03:35]

<nick_fn>

with a html link to the actual diff in svn

[03:35]

hopefully in bugzilla there's a way to not pollute the comments

[03:36]

<nick_fn>

we see this as *part* of the comments

[03:36]

hmm, i much prefer like they do down at xmms2 bug tracker. iirc they have a special field for commit references

[03:37]

<anders_>

We (xmms2) have both. A custom field for commit-id. And a comment added when change referring that bug automatically added when changes is pushed to trunk (xmms2-devel.git)

[03:40]

<nick_fn>

can I tell (from a shell) what my current firmware is?

[03:40]

i think if we do it, we'll be using a custom field. testing team really thinks in terms of releases and is not big on having too much stuff in comments. confuse them a bit often

[03:41]

nick_fn: you can't, AFAIK

[03:41]

<nick_fn>

it would have to be a custom field *list*, because you can obviously have more than one commit per bug

[03:41]

nick_fn: that's true

<nick_fn>

plus then you don't see them in the correct "timeline", but maybe that's just me :-)

nick_fn: or commit numbers separated by spaces

nick_fn: timeline is enforced by hook

<nick_fn>

what about seeing the commit messages from the bug page?

[03:43]

you go to svn

[03:43]

<nick_fn>

by timeline, I mean we have "regular comments" and "commit messages" in the same list on the page, so you can see the order of history properly

<nick_fn>

nerochiaro: that's not quite as nice - if you're looking for something in particular, you'd have to click *each* commit ID, rather than just reading the comments and seeing which you want.

it would be ok for me and you, but make no sense at all for testing team

<nick_fn>

ok

i know it's nicer your way, no doubt about that

<nick_fn>

Will Arizona have sshd at last?

no one worked on that officially, sorry

bugzilla.neurostechnology.com

<nick_fn>

do we have all pieces to run it yet ? last i checked we missed random generator that was good enough

<nick_fn>

no idea

<nick_fn>

I just know I have to keep coming here for a password, rather than using an ssh key ;-)

we never change it

it's the same since we first added it

<nick_fn>

yeah I know, but I keep having to type it :-P

[03:48]

<nick_fn>

and scp doesn't work, etc.

[03:48]

fair enough :)

[03:48]

<nick_fn>

no matter, I'll wait patiently

[03:49]

IIRC we can have ssh, but it would be as secure as telnet due to some lack in the random number generator that basically defeats the purpose of ssh encryption

[03:49]

but i checked a lot of time ago and never in depth

[03:55]

<crweb>

typically security is a pain in the ass

[03:55]

<crweb>

which is why its such a global problem

[07:20]

nick_fn: hey, you can get firmware version via shell with "ubootcmd get cur_rootfs"

[07:55]

<nick_fn>

thanks greyback

[07:57]

nick_fn: np

[10:49]

<orion-uk>

hay Greyabck :) thanx 4 the post about the audio decoder :) any idea what it could be or what I can try?

[10:50]

<orion-uk>

set a scedualed recording last night 4 a film and it recorded 1 min 53 secs only!

[10:52]

orion-uk: to which media were you recording ?

[10:53]

<orion-uk>

nerochiaro: Hi, usb hd.. 7200rpm

[10:53]

orion-uk: self-powered ?

[10:54]

<MattJ>

orion-uk: Poor you, I 9 minutes out of 60 :)

[10:54]

<MattJ>

*I got

[10:55]

<orion-uk>

size 150gb, seperate power cable yes

[10:55]

<orion-uk>

hi Mattj ;-)

[10:55]

<orion-uk>

prob still wiv u 2 then Mattj :-/

<MattJ>

Yup

<orion-uk>

ggrrr :(

well, both of you, unfortunately i need a log from console to be able to make sense of what's happening (or try to at least)

<MattJ>

Though I still can't rid the suspicion that it is related to free space

MattJ: it's possible

forums.neurostechnology.com

<orion-uk>

<MattJ>

But it happens randomly

[10:57]

<orion-uk>

;-)

[10:57]

<orion-uk>

same here.. but more than not!

[10:58]

<orion-uk>

112gig free btw

[10:58]

<orion-uk>

hows tricks MattJ?

* MattJ <-- been very busy

<orion-uk>

good&bad thing then

<MattJ>

Heh, yes :)

<orion-uk>

:-)

[11:00]

<orion-uk>

me 2, got a wks agency work at a construction site 2 get me by... darn wind n rain ! ;-/

[11:00]

<orion-uk>

gone back 2 my old roots 4 money untill some better job comes along.

[11:01]

<orion-uk>

nerochiaro: did that link help at all?

[11:01]

sorry, looking now

[11:01]

<orion-uk>

np, k, ty.

orion-uk: no, that's not what i need, that's a dmesg log. what i need is simply to connect to the serial terminal and log to a file the stuff that the OSD spits out there (before and during recording)

greyback: regarding what you wrote there, the dmesg log and serial log aren't the same

greyback: dmesg is generally only from kernel

nerochiaro: ahh, fair enough, my bad :(

<orion-uk>

ty 4 ur input though greyback :)

I figured since the codecs were via kernel modules that's where their output would go

<orion-uk>

nerochiaro: not possible 2 do serial link 4 me :(

orion-uk: no prob, sorry for wasting your time!

greyback: the bulk of what is going wrong is in nmsd, which is userspace

orion-uk: well, thanks anyway for reporting, it's useful to be aware of the issue

<orion-uk>

has.. yuor times more valuable than mine atm..thanx 4 giving it :>

something fishy is going on in these recent firmwares on recording, unfortunately i don't know yet what

<orion-uk>

so not much i can do my end (for now) then?

<orion-uk>

k

nerochiaro: hmm, makes sense. Pity serial is the only resort

orion-uk: keep an eye on how often that happens

greyback: you may be able to turn on syslog

<orion-uk>

am on 3.33-1.73-04.81t

[11:07]

<orion-uk>

pretty much every recording tbh.. strange thing is...

[11:07]

try checkign the wiki, there should be a page about it. or the ML. i used it only once and wasn't the one who put it there, but IIRC it sould still be there

[11:08]

<orion-uk>

if i have a few recordings (say 4-5-6) it starts to work ok, if i remove the hd to store them on my other pc or delete any OR upgrade fw,unplug HD.. it goes "funny" again!

[11:08]

nerochiaro: it'd be worth getting going. I'll have a look

orion-uk: 4-5-6 already completed recordings, or scheduled ?

<orion-uk>

pretty much a gamble on any recording working tbh

<orion-uk>

...

greyback: thanks

<orion-uk>

it happend on either

[11:10]

<orion-uk>

if i have 4-5-6 recordings (sucsessful or not) then it seems to start working fine!

[11:10]

<orion-uk>

untill i do any of the above.

orion-uk: sorry, i'm not sure i understand

orion-uk: where do you have these recordings =

?

<orion-uk>

sorry am i not explaining properly..

or i'm thick ;)

<orion-uk>

recordings on usb hd

<orion-uk>

;-)..lol nah

<orion-uk>

had a long day, prob not got my head on..lol

so if you previously recorded from 4 to 6 movies, it works, if you remove these, it doesn't ?

[11:12]

<orion-uk>

yah

[11:12]

<orion-uk>

but this is not always the case.. just a pattern i seem to see, but could mean nothing!

it actually sounds pretty weird, but i've seen stranger things

<orion-uk>

lol

<orion-uk>

maybe im putting 2x2=7..lol

well, let's see if anything comes up with that log, in case anyone manages to get one

greyback: thanks for giving it a shot

[11:14]

also, if you folks downgrade to something older, does it start to work again properly, i assume ?

[11:14]

<orion-uk>

have tried all fws avaiulable with no diff... beta,dev,officials...

[11:15]

that's strange, since this rate of failure is something we didn't have in the past

[11:15]

with past firmwares, i mean

[11:15]

<orion-uk>

only real change i saw was the youtube sertver updated.

orion-uk: well, that was pretty recent

i mean downgrade, not upgrade

go back to older firmwares

<orion-uk>

i have only had my osd a couple of wks...

[11:17]

<orion-uk>

so i should dl old fws from website instead n try them?

[11:17]

well, that may help until we get a fix on the issue

[11:18]

<orion-uk>

ok.. will look on w-site 4 1 now n try

[11:18]

<orion-uk>

any recomendations on how far back 2 go?

[11:18]

go back in steps until you find one that works, and then let us know what it is

[11:19]

i mean, one step at a time

[11:20]

<orion-uk>

ok, will keep u posted... funny thing is.. when i tried to re-produce the error for greyback.. it would not happen..lol

[11:20]

<orion-uk>

will try some fws and get back 2 u ok... thanx for the advice ;-)]

[11:23]

<orion-uk>

even if i loose youtube by downgrading atleast my recordings will work (i hope) lol...well better let u get on for now.. thanx again Nerochiaro, MattJ,Greyback :-))

[11:23]

orion-uk: good luck!

[11:23]

<orion-uk>

greyback: ty :)

[11:24]

nerochiaro: strange, syslogd refuses to run, bailing with 'readonly filesystem' even when I give it a path to a rw dir

[11:25]

<orion-uk>

anyone: could a hd overheating be a cause? (maybe i could put a fan on it!) does feel a lil too warm at times.. but, i have had this prob even when using other usb hds.

[11:27]

orion-uk: I doubt it, it's usual they get a quite warm, just put it on a surface where air can circulate

[11:28]

<orion-uk>

is out of case sitting on foam atm. anyways...am off for dinner, thanx again all... cath u l8trs :)

[11:34]

<MattJ>

nerochiaro: my firmware isn't very recent - a beta from some weeks before Christmas

[11:37]

sorry, got a phone call

[11:38]

MattJ: i'm not sure why syslog bails. i don't have a firmware installed of that version you're running to check

[11:38]

but wait a second, i'll try

[11:41]

MattJ: it's trying to access /dev/log , not sure why

why it's missing, i mean

<MattJ>

syslogd?

ya

<MattJ>

It's an alternative to the serial console then?

[11:43]

most applications log with a system that goes both to syslog and serial. kernel goes to dmesg and/or serial. badly written apps to serial only

[11:43]

<MattJ>

k

[11:43]

<MattJ>

I would really like the serial to be configurable

<MattJ>

Enable/disable, baud, etc.

MattJ: i think it's something the hardware doesn't let you do, but not sure

<MattJ>

I see

<MattJ>

I asked on the ml a long time ago

<MattJ>

Someone sent instructions on how to change it

[11:45]

<MattJ>

But it involves rebuilding the firmware, since there is a call there to open the serial port

[11:45]

<MattJ>

and also set params

[11:46]

isn't it something you have to pass as a parameter to kernel ?

[11:47]

and anyway, uboot won't be able to be configured very easily anyway

[11:48]

well, actually, via ubootcmd, maybe

[11:48]

on second thought it may be something interesting to check

anyway, to get syslogd running, you need to create that /dev/log device, which sadly you can't do since it's ro fs

:(

that needs to be fixed clearly

<MattJ>

groups.google.co.uk

[11:53]

MattJ: you can change that argument via telnet, using ubootcmd

[11:54]

<MattJ>

Yes, but won't help if I don't rebuild the code?

[11:54]

<MattJ>

Oops, gtg, bbl

[11:55]

it will. you will not be able to attach console to uboot, but you will be able to attach it to kernel. the parameter you need to change is called "console" and the value is "console=ttyS0,115200n8" (you need the "console=" part in the value, it's not a typo)

[12:40]

how hard would it be to support other video sites besides youtube?

[12:41]

antics: i posted about what that would involve in mailing list recently

[12:41]

nerochiaro: thanks, i'll go look thru my mail (sadly i rarley read it)

[12:42]

antics: no problem. you can also search via the google groups web interface. maybe you can find it faster that way

[12:43]

nerochiaro: there have been tons of firmware updates lately, when has the release cycle gotten so short?

[12:43]

not that i'm complaining mind you, just curious

[12:44]

antics: define "lately"

[12:44]

;)

[12:44]

heh, i'd say the last few months

up until december (i think) we went with cycle of a release every three weeks or so (for officials)

I'll guess i've had mine for 6mo, in the first ~4mo there was 1 update

oh, thats about right

best feature is remembering samba credentials

[12:46]

antics: of course if you moved your auto update settings to the [dev] channel, you got almost daily releases :)

[12:47]

antics: right now (last months) we are releasing very very slowly since we're working on a completely new line of software based on QTopia

aren't there royalitys with that?

not anymore since quite some time

qtopia-core at least