Log from #neuros IRC channel

i see, then it might be feasible

<crweb>

a $8 universal remote will control just about any dvd/tv/vcr

<crweb>

my phillips tv is 0018

<crweb>

my vcr is 1002

[02:42]

<crweb>

you just have the osd listen for say.. Sony VCR model zx100f

[02:42]

<crweb>

then you use the motorolla ir setup to "browse" for codes that work right

[02:42]

i'm sure something along the lines of the ir blaster learning can be done on the OSD to accomodate choosing which remote code to listen to

altough might need modifications to the MSP code again

<crweb>

well, blaster learnind would be different that IR listening

<crweb>

oh wait

<crweb>

scratchthat.

<crweb>

it would be better to set the osd as 1 device

[02:44]

<crweb>

and use the cable box's universal remote setup

[02:44]

<crweb>

they don't have to learn codes. They bought licenses for the IR databases

[02:45]

<crweb>

very interesting idea i think

[02:45]

<crweb>

let the cable box do program guide/scheduling..

[02:45]

<crweb>

(not that the osd shouldn't do it too)

[02:47]

crweb: we need to learn codes, we need to know 1) which remote brand they will blast to us and 2) which key on that remote should trigger our recording

[02:47]

<crweb>

no no, the user chooses the brand they blast

<crweb>

we tell them to try "sony remotes"

i know, but how can the osd possibly know what the user chooses on the box ?

<crweb>

it doesn't need to know

why not ? if it should listen, it should know

<crweb>

the osd listens for specific sony remote

<crweb>

the user picks, try sony remote 0015

crweb: aaaah, ok. sorry for being that thick

<crweb>

if that doesn't work try sony remote 0018

i get it now

ok, it's a feature to keep in mind. but i think our own remote guide comes first

[02:50]

<crweb>

well, this simply would be as easy as stop/record ir code that osdmain/recording system listens for and stops/starts recording

[02:50]

<crweb>

you are right, but not to complicated

[02:51]

yep, if the box can support that, using just the neuros record key code would be the best. anyway i'm going to have some breakfast now and do some work for the old folks, they're still paying my bills for now ;)

[02:51]

<crweb>

set vcr control mode to dump recordings to <dir>, put osd in controlled mode

[02:52]

<crweb>

it would be best.. but those codes seem to be some other format

[02:52]

<crweb>

have a good day

[02:52]

and you have a good sleep

[03:06]

anders_: ping ?

<anders_>

nerochiaro: pong

anders_: just wanted to know if the xmms2 section in the SoC page is ok for you

anders_: and if you're ok to be listed as mentor then

for that project

[03:09]

<anders_>

nerochiaro: Looks fine; I also liked it from here: wiki.xmms2.xmms.se

[03:10]

anders_: what is your google account ?

[03:10]

<anders_>

andersg@0x63.nu

[03:12]

anders_: ok. and linked back from our SoC wiki page to yours :)

[05:37]

hello

[05:38]

GuestN398: hi

[05:39]

I'm a newbie triying to hack the Neuros OSD

[05:39]

can you help me?

[05:39]

sure, i can try. what do you need ?

[05:40]

I have installed the Virtual Machine

but I don't kwon the root password

"neuros2"

user "neuros"

user is neuros and pass neuros2

but what about root ??

if you need root just use sudo

sudo command_you_want_to_run_as_root

It ask for a password

insert the user's password

I left empty and it does nothing

ok

It works !!

:)

Thanks a lot !!

no, problem. can you please pick a nickname the next time you login ?

so if you return later we know you are the same person

ok

I will

thanks. and welcome :)

open.neurostechnology.com

grazie ;)

[05:45]

hehe, di nulla

[05:49]

<suma>

for long time, no release in firmware !

[05:50]

suma: i think they're about to release a new one. there's a new announcement in odnt submission queue

[05:51]

suma: this will have much graphics redesign, it seems

[05:57]

<suma>

nerochiaro: any link for that to download ?

[05:57]

<suma>

thanks for the info

[05:58]

use at your own risk

hey nero

daurnimator: hey

hows life?

suma: the OSD will already pull that from the internet if you use the firmware upgrade feature in osdmain

[06:00]

<suma>

i c

[06:00]

daurnimator: all fine, finishing the last things for the old employer

[06:00]

<suma>

let me check it

[06:01]

suma: i'll do that too. i really have to kick the release folks though, they should not release for network upgrade without an OdNT official announcement

[06:07]

nerochiaro: anything i should boot up my OSD for?

[06:25]

<suma>

nerochiaro: got updated, only the UI is changed, still i could not play the xvid movies

[06:26]

<suma>

hope to see better movies as soon as you join neuros

suma: i can't do much abou that. it depends on the codecs, and the codecs depend on Ingenient

sorry

and i reall don't like the new UI. it's too dark, i have to ramp up the lightness of the TV to see it properly

<Ycros>

it's so frustrating

Ycros: i hate that too

[06:46]

<Ycros>

hmmmmmmm

[06:46]

<Ycros>

are there other companies like ingenient that provide similar sort of stuff?

[06:50]

i think there's another calle ITTIAM or something like that

[06:50]

seems better from what other people said, but it's all hearsay

[06:50]

<Ycros>

looks like www.ti.com has a list or something

[06:51]

but at this point, we can't just more or mix, unless we want to start from scratch on some other vendor

[06:53]

<Ycros>

oh, I agree

[06:54]

<Ycros>

it's just that I would seriously think about using them again in the future

[06:56]

well, you have a point, although there might be also other factors on the table i don't know (price, politics, estabilished relationships, that stuff)

[06:57]

<Ycros>

"it doesn't work properly, and they haven't fixed it" is a pretty big factor that I can see

[06:57]

<Ycros>

but yeah

[06:57]

<Ycros>

I dunno

[07:04]

Ycros: we'll see when the time comes

[07:51]

<anders_>

nerochiaro: Address space is limited on arm?

[07:58]

anders_: actually i'm not sure about that, sorry

[07:59]

<anders_>

Isn't it stardard 1:3G split?

[08:00]

anders_: i really can't say

[08:02]

JoshMalone: maybe you can help on this, without having to look it up ?

[08:03]

?

[08:03]

what's up

[08:04]

oh, whoops, anders wanted to know if the address space of our ARM processor has some quirks or is limited somehow

[08:05]

erm....afraid I don't know the TI arm very well

[08:05]

<anders_>

I was referrirng to "This will help in achieving suitable performance, but care must be taken not to exceed the comparatively limited address space in the ARM."

[08:05]

every arm has it's weird physical address mappings

[08:06]

on the Intel chips you just had to go to the book

[08:06]

<anders_>

No matter how things are mapped each process has 3G _address space_ on ARM, no?

[08:07]

afraid I don't know the answer to that. sorry :(

[08:08]

honestly, most of what I did on ARM boards was concerned with flash storage and a bit with GFX chips

[08:08]

<anders_>

On other (32bit) architectures linux uses 3:1 split at least. 3G address space for userspace, 1G address space for kernelspace.

[08:11]

anders_: that line was mine... i thought i pulled that fact from within the kernel docs itself,

[08:12]

<anders_>

srobertson: Oh..

[08:12]

but it could be that I was mistaking the ARM's physical addressing limits with its virtual addressing limits

[08:12]

www.samsung.com

[08:13]

lists a 926EJ as having an "Address space : 512MB"

[08:13]

<anders_>

Guess that referrs to physical address space.

[08:14]

what possible cause can there be for limiting the virtual address space ?

[08:14]

nerochiaro: no MMU on chip

[08:15]

that's not saying the 926EJ doesn't have one, just that if it didn't, the physical and virtual could have to be the same

[08:15]

* srobertson is slightly out of his league here

[08:16]

same here, it was an honest question, not a rethorical one

[08:16]

<anders_>

Yeah, without an MMU there wouldn't be much virtual address space to talk about :)

[08:18]

yup, afaict i made that line up

[08:19]

Hi. I'm making a package builder "ybuild", and a pakcage prune system "crosspm" along with a package manage, for OSD. And I'm considering how to merge packages when inserting removable-storage. I plan to use a daemon "cpkgd" to handle the removable-device event. Any other idea, or existing resolution?

[08:20]

hellwolf: there's hotplug that already manages removable storage, no ?

[08:21]

Yes, I know. I need to handle the event it generates also

[08:21]

hellwolf: you can add your scripts to the existing hotplug script, like i did with unionfs

[08:21]

hellwolf: what would you use the deaemon for ?

[08:22]

To check dependencies. And only add softwares that met the deps

[08:23]

There's a db on removable device, and when it's inserted, cpkd trying to merge it to system wide db

[08:25]

ok, but why do you need a daemon ? can't you just trigger that from inside the hotplug script ? why do you need to have a daemon always running ?

[08:29]

Ok, you're right. I just need some trigger scripts then.

[08:30]

hellwolf: also be careful, because someone can insert a media card and just immediately after another media card in the other slot. if your central database is still updating something bad might happen ?

[08:32]

That's possible. I could use a lock file to protect the db, right?

[08:33]

hellwolf: something like that, or a linux semaphore

[08:38]

That's it. And another problem is, since maybe I can not import all the packages from removable media(if some software do not meet all the deps), unionfs is not sufficient then. Is the "symlink farm" really a bad idea, or to make a special FUSE filesystem?

[08:41]

hellwolf: well, some suspect FUSE might be too heavyweight for the OSD, but you can try investigating it. as for the symlink farm, i've never been a fan of symlinks, but if it works well, why not ?

[08:41]

<anders_>

What do you mean with "symlink farm"?

[08:41]

* anders_ likes encap symlinks..

[08:42]

anders_: that's how i interpreted it

[08:43]

anders_, an interesting software called "stow" makes "symlink farm" for you ;)

[08:43]

<anders_>

hellwolf: Sounds exactly like encap :)

[08:44]

they use the same idea i think

[08:44]

it's also said in encap page that they took inspiration from stow

[08:45]

I haven't use FUSE on arm before. I want to try it to see if it's bearable for embed system.

[08:47]

FUSE is interesting

[08:48]

hellwolf: if it work well, how will you use it for the packages ?

[08:50]

I want there's something like /opt which is controled by FUSE or symlink farm, according to the package database

[08:50]

The softwares on removable-media is all compiled with prefix "/opt"

[08:51]

They will be shiped with special binary packages like *.cpk. User installs it, and record it in database in removable devices

[08:54]

<anders_>

And FUSE will be used to look inside the cpk-files I assume?

[08:56]

or will it be used to make only part of these packages (only the parts that satisfy dependencies) look like they are part of the OSD file system ?

[08:56]

that is, look like they are "installed"

[08:58]

Yes, hopes I haven't make things too complex. But it's close to midnight here, thantks nerochiaro anders_ , see you all

[08:58]

hellwolf: will you join the summer of code if we get approved ?

[08:59]

nerochiaro, sure, that's interesting

cool

:) see you

bye

<anders_>

hellwolf: The concept is pretty nice.. Don't know how well FUSE works in low memory environments..

[09:01]

hellwolf: you should really post some document to explain that better, after you have investigated if FUSE works ok

[09:01]

<anders_>

Data will take double space in the cache I guess.

[09:01]

anders_: it's sure more overhead than symlinks

[09:02]

<anders_>

Indeed. It is nice to be able to just put single package files on the card though.

[09:02]

anders_: you can still do single packages and work with symlinks

[09:02]

<anders_>

instead of huge directory trees. Which _will_ screw up when copied to memorycards from windows..

[09:03]

<anders_>

nerochiaro: How?

[09:04]

well, actually it's just an idea i had on the fly ;) making the package a disk image and loop mounting it

[09:04]

a file system image, i mean

[09:04]

or just cramfs or similar

[09:05]

<anders_>

Right...

[09:05]

<anders_>

Wonder if that has the same diskcache interaction as a FUSE..

[09:05]

yeah, it's all a matter of investigating a bit.

[09:06]

<anders_>

A compressed filesystem would probably have..

[09:06]

<anders_>

Maybe an uncompressed can reuse the same disk pages..

[09:07]

i don't actually see anything against having the uncompressed disk image. the disk image would only be used to work around the media card "host" FS problem anyway

[09:07]

and for convenience

[09:14]

<anders_>

Yes. A simple ext2 image would work like a charm.

[10:24]

nerochiaro: ping

[10:24]

may1937: hey, finally we are online at the same time ;)

[10:25]

may1937: i just wanted to know if you are interested in being a SoC mentor for neuros

[10:25]

yeah heh

[10:25]

and if you want to add some DSP-related project for SoC

[10:26]

yes crweb mentioned it last night, i am interested

[10:26]

and even managed to get srobertson interested in applying

[10:27]

srobertson as student, i suppose ?

[10:27]

aye

[10:27]

just saw your ml message, i can add something to the wiki

[10:28]

ok, students will apply later, if we are approved, so there's time. but if you want to help as mentor, i need to have your google account and possibly real name

[10:28]

and yes, if you can add something on that wiki page, that would rock

[10:30]

you need a google account? that seems weird

[10:30]

they need it. each mentor needs to have a google account (it's not gmail, just google)

[10:31]

unless my eyes crossed at some point while i was preparing the SoC paperwork yesterday night

[10:34]

so is it enough to have the wiki done by the 12th, or do we need a more "serious" proposal?

[10:36]

ok have created account with that other email addy

[10:40]

<may1937> so is it enough to have the wiki done by the 12th, or do we need a more "serious" proposal?

[10:40]

<may1937> ok have created account with that other email addy

[10:44]

added at least a start to soc project page, need to get going to work now

[11:34]

may1937: sorry, i had some network problems earlier

[11:34]

i'll add you as mentor with that other account.

[11:35]

may1937: the project entry is also good. thanks

[11:36]

may1937: do we wanna think about cgen as a potential project too?

[11:36]

what is that ?

[11:45]

over in #gcc-c54x we (may1937 and I) are pretending to use cgen to generate a machine description of the c54x for cgen, which will supposedly give us binutils and an emulator, as well as a starting point for a gcc backend

[11:45]

when I say 'pretending' I mean we're in need of motivation

[11:47]

how valuable would an open-source c54x compiler be to Neuros and the world at large?

* srobertson hasn't quite had the idealism beat out of him yet

personally, I'd be more interested in c64x, that's the one on davinci, right?

mainly the ability to write code for the dsp without spending $lots on a compiler

actually TI as this event said they'd be willing to give a free version of the compiler for c54x

well then

it wouldnt be gcc or open source and I know not satisfy the purists,

this was concession made to us for our community believe it or not

that is cool

they are really enthusiastic about building a community around their davinci stuff

no way am I *that* idealist

oh way, the c54x compiler is still windows only and wine is messy

I really need to run though, I think some open source tools for the davinci could be on the list.

KVM + new Core 2 Duo on the way... wfm

we might even collaborate with fsf on that project.

JoeBorn: that wouldn't be SoC material, tho, right?

i mean if it is, SUPER awesome, but i'm guessing no.

srobertson: the compiler project?

JoeBorn: you'd kinda have to if you want the gcc backend to be part of the gcc tree

You have to assign your copyright to the FSF to have code accepted to the project

no, unless you mean c64x, but still that'd be nuts. just c64x stuff in general.

well, maybe we can break it in a piece that'd be reasonable?

I don't know enough about it

chreekat: right, but I mean even co-working on it as a project

or co promotion or co something