Log from #neuros IRC channel

[04:13]

JoeBorn: you there ?

[04:13]

yep, I'm here

[04:15]

I know I should know this, but when's your last day again?

[04:16]

Is there any hardware jack that could make a Wii like remote controller, I think that would be cool

[04:16]

I was sorting through these long emails i suppose you wrote on the plane to china. can you elaborate a bit on the one about the 4 months integration, and the point about the need to refactor the software. the last day will be this friday.

[04:16]

hellwolf: isn't that a bluetooth controller ?

[04:17]

hellwolf: maybe a BT dongle and some sw driver can do that

[04:18]

I don't care if it's usb or bluetooth, I think a Wii-ish controller could be used as a mouse.

[04:18]

BT dongle?

[04:19]

hellwolf: bluetooth dongle. you plug that into USB port and it does BT

[04:19]

ok , I googled it

[04:19]

That's cool

[04:20]

hellwolf: and good idea for the wii-like mouse controller. but nintendo would sue us ?

[04:21]

nerochiaro: I have the email open.

[04:21]

well, was the rest of the email clear?

[04:21]

my point was that we should be able to do a 6 month porting/integration

[04:21]

JoeBorn: it's basically about the timing in going on a new hw platform with current software, do i get the general outline right ?

[04:22]

yes, that's right

[04:22]

and my feeling is that one serious threat to that integration will be if our software is a "mess"

[04:22]

I don't know if there's patent issue, but wii-like controller attract me much

[04:23]

ie. if the interface is not clean then we'll spend tons of time debugging

[04:23]

hellwolf: as a mouse for what application?

[04:23]

well what did nintendo do that's unique

[04:24]

there have been motion controllers around for a long time, they didn't invent that

[04:24]

and ther've been gyroscopic mice and such for a while.

[04:24]

JoeBorn, such as OSD, we can use that controler to input something on screen by some supported software

[04:24]

hellwolf: I see.

[04:24]

well, it's certainly worth some discussion

[04:25]

I haven't played with a wii

[04:25]

I suppose I should

[04:25]

Input "abcd" by "1234" is horrible feeling

[04:25]

hellwolf: well new controller or not, we'll fix that

[04:26]

JoeBorn: well, yes and no. the fact is that the application is only dependent on a few hw-specific things, and most are abstracted away in cooler or the OS. if we continue down that path further, like discussed in refact. meeting, it's not going to be such a nightmare.

[04:26]

That's cool.

[04:26]

JoeBorn: would help to have a semi-automted test suite for hw-specific parts of cooler, maybe

[04:27]

JoeBorn: so we can quickly see what gets broken without having to actually use the application

[04:27]

JoeBorn: i mean, osdmain

[04:27]

nerochiaro: as you know, I don't have a very clear picture of the whole architecture, refactoring discussion, but I just have a vague feeling that we have these unattached strings that link various parts of the code in a non abstracted way

[04:27]

maybe I'm wrong

[04:28]

nerochiaro: agreed

[04:28]

*if* we do the refactoring and keep it "refactored" then it shouldn't be such a nightmare

[04:28]

and 6 months should be reasonable.

[04:29]

I find it impossible to believe that all these guys cannot deliver a working framework in 4 months, this is what they do, day in and day out

[04:29]

JoeBorn: i think that yes, it can be a reasonable time, barring issues with vendors in stuff we don't control

[04:29]

as long as our application code is not some hacked together steaming pile, we should be ok.

[04:30]

well, I'm not saying every file works after 6 months, I'm saying that we have a working, gamma level product that we continue to evolve...

[04:30]

but I think we're on the same page...

[04:31]

gao keeps saying it's a 12 month integration and I just can't accept that.

[04:31]

he's not really referring to a specific path

[04:31]

and he acknowledges that others do it in 6 months, but....

[04:32]

What OSD is using for ethernet and usb controllers, some IP blocks in the ARM processor (with dediceted drivers) or some "standalone" chips,with standard drivers?

[04:32]

it took us well over 12 months last time.

[04:32]

*dedicated, sorry

[04:32]

mattepiu: AFAIK, the drivers are reasonably standard

[04:33]

the USB host is built into the DM320, but I gather the USB code came from the standard linux tree

[04:33]

However IANAH

[04:34]

IANAH is the reseller of the USB host core IP? Thank you

[04:35]

IANAH was a joke, (I am not a Hacker)

[04:36]

mattepiu: the usb code is open source, but the drivers don't work completely, something is not implemented and prevents USB HID to work correctly

[04:36]

nerochiaro: and my point on the vendor side was, we are not experts on that area, we should hand it over to them and hold them to high level deliverables

[04:36]

nerochiaro: ok, then is probably the opencore solution

[04:37]

rather than us flailing around and participating in their av stuff

[04:37]

mattepiu: they are in the kernel, you can check that in SVN if you want, i don't remember right now where exactly

[04:38]

JoeBorn: the codecs especially should be 100% vendor responsibility. unless they're open, they should have them working themselves, and in the time frame they specified at contract. it seems pretty standard to me, no ?

[04:39]

yes, but I think we should even go a level higher.

[04:39]

and include the parser and framework, etc.

[04:39]

any sync issues etc are their issue.

[04:40]

we provide them a test suite of files and they are responsible for playing them properly on their complete framework

[04:40]

with parser you mean the header parser, to recognize file format ?

[04:41]

their complete framework allows us to pass them some files and the output is a properly playing file that can be rewound, fforwarded, etc

[04:42]

well, I don't really know what I mean, just that file --> vendor black box --> proper playback

[04:42]

if not, we sent it back and say "doesn't work" not some horseing around where we do a stuck pig imitation, etc

[04:43]

and we make this high level instruction very clear to them so that they can quickly solve problems internally before releasing to us.

[04:43]

at least that's my idea

[04:45]

JoeBorn: well, 2 points. first, parsing headers ourselves is something we need to do anyway for metadata (especially in audio files), so it's ok if they do it too, but i think we need also. 2. an interface that says "here's a file, play it" is fine and dandy. however it's not enough, because that won't work for streaming, transcoding, and stuff that genereally don't have a real file. being able to send pieces of data to the codec w/out file

[04:46]

the "here's a file, play it" however is good, and is the best way to have them hold up to their part of the bargain

[04:46]

i'm just saying we need more than that

[04:46]

well that's fair enough, I guess my point is let's spend more time on the high level spec so that the instructions to them are clear and it will give them the ability to troubleshoot their own stuff

[04:46]

oh, and that "oddball formats" comment from Ingenient rep. was quite a faux step on their part, if you ask me

[04:47]

and shorten the feedback loops

[04:47]

agreed

[04:47]

if they have any interest in effectively solving the problem, they will at least have the release procedure that allows them to quickly test and go back to their team if there's a problem

[04:47]

rather than all of us flailing around (see above stuck pig imitation)

[04:48]

i'm not even entirely clear right now what's the procedure to get vendor stuff fixed, if any

[04:49]

i.e. what's the current stuck pig dance with vendor

[04:49]

nerochiaro: well, there's some truth to that, but I think the situation could have been significantly aided by better planning on our part

[04:49]

nerochiaro: that's my point, if you have watched a stuck pig flop around, then you probably know he doesn't follow a set dance pattern :)

[04:50]

surely we need a format test suite and then a semi-automated test harness (that at least tests all formats in the suite for crashes, of course a human is needed to judge picture glitches or such things)

[04:51]

also some way of shortening to the minimum all clips we get while preserving the issue

[04:51]

and no, thankfully i have not ever watched stuck pigs. but i get the idea :)

[04:52]

I'm fairly sure if you search youtube, you'll find something if needed.

[04:53]

oh i'm sure. but i won't

[04:58]

JoeBorn: as for the ittiam v/s ingenient thing, i have no idea who's worse of better. from what you say ittiam attitude seems much more serious and less "cranky", for lack of a better term. i mean, when i picture ingenient in my mind for some reason i have this picture of a room full of old soviet beaurocrats and stacks of dusty paperwork (surely i'm just over-imaginative, but still)

[05:00]

oh, and another thing. if we make the media part that is in the hands of a vendor too "high level" don't we end up with the risk of closing up too many stuff ?

[05:01]

well, I think it is important for us to have access to be able to fix/modify stuff over time

[05:01]

my high level comment is really directed towards how we manage the deliverables.

[05:02]

spend more time up front and less time flailing down the road...

[05:02]

that's really what I'm saying I guess

[05:02]

oh, that makes sense then

[05:02]

and don't flail along with the vendor, just hold them accountable to the agreed deliverables

[05:02]

i was just making sure we don't erode too much the open part of this joint

[05:02]

Joe: Sorry, been busy lately. Gonna download the new firmware and do a little testing for you today. Don't have enought time to really get going on the english stuff yet though.

[05:03]

theolodian: english stuff ?

[05:03]

nero: a while ago he asked for help from us lackeys on grammer

[05:04]

theolodian: thanks, of course whatever you can do will be appreciated..

[05:04]

JoeBorn: by the way, should't some stuff from these emails be published, so people know what we talked about instead of guessing it from refences ? ;)

[05:04]

sure, I suppose so.

[05:04]

Joe: I also appreciate your current discussion. I have engineers arguing about whether it will take 6 or 12 months to port over to ARM, and if suppliers can assist us or not. :|

[05:05]

theolodian: ah, grammar in the firmware ? i corrected some stuff like that last time i dived in, but i'm sure there's more

[05:05]

nero: In the interface

[05:06]

Would help if I could spell grammar correctly though . . .

[05:09]

did i spell it wrong ?

[05:10]

i know i probably got wrong the past of "dive"

[05:10]

Crap, now you're going to make me check. I thought I spelled it wrong as soon as I posted it.

[05:10]

theolodian: being a grammar/spelling nazi has that downside. you keep double checking yourself ;)

[05:11]

thankfully i am not one (and probably couldn't ever be)

[05:11]

Yes. And yes, it is grammar.

[05:12]

DIdn't mean to kill your chat with Joe.

[05:12]

theolodian: you did not

[05:13]

OK, cool

[05:30]

<crweb>

JoeBorn: keeping it refactored i think may be a chore

[05:30]

<crweb>

JoeBorn: old habits die hard

[05:31]

crweb: agreed. I'm counting on you guys for that.

[05:33]

I do believe in my heart that its crucial and I'm all ears for whatever I need to do to keep it clean

[05:33]

<crweb>

I agree that while highlevel stuff as in "Play file" is good, we have to have the ability to play non file formats

[05:33]

<crweb>

it is a must.

[05:34]

agreed.

[05:34]

<crweb>

if they are any sort of designers at ingen. they shouldn't need to be provided such a high level of abstraction to understand the goal

[05:35]

<crweb>

really, their codecs should provide "play this stream"

[05:35]

play file was probably a poor choice of words

[05:35]

<crweb>

and then we provide the stream

[05:35]

<crweb>

we can stream a file, or a stream

[05:36]

well, my point simply is that who ever it is, we need to present them a test suite of files and usage, etc.

[05:36]

<crweb>

right

[05:36]

whether or not we should or should not *have* to do that is academic

[05:36]

there is little doubt that in practice we do

[05:37]

<crweb>

yeah.. i was about to say, why are we providing the designers with the tools to test their own code..

[05:37]

<crweb>

they don't have many customers do they?

[05:38]

no, they have lots, but grab a random file and try to play it on a random device, you won't have much luck

[05:38]

this codec/wrapper business is the bane of the digital video business

[05:38]

<crweb>

still.. they should have some sort of core interface that sits on top of the codecs

[05:39]

well they do have that.

[05:40]

<crweb>

kinda odd... it all seems kinda strange to me. Maybe if I heard their explinations that would effect my interpretation

[05:40]

<crweb>

i myself am learning how to program for the dsp right now as well.

[05:40]

<crweb>

maybe i'll understand better as i get more familiar

[05:41]

<crweb>

i got caught up reading a david eddings book and stayed up to late. Have a good night/day/afternoon.

nite

wow, night

crweb: wait

<crweb>

yeah

[05:47]

<crweb>

i'll have the firmware updated in the afternoon

[05:47]

quick question, this free dsp compiler from TI, do we want for linux host or PC

[05:47]

<crweb>

theres a free dsp compiler?

[05:48]

free as in beer, and yes they offered it at the dev conference

[05:48]

I guess they are going to take a snapshot and give us some kind of limited version

[05:48]

<crweb>

sweet, anyway i can get my hands on it?

[05:48]

if it's just "as in beer", migth as well be for windows and we use wine. although a linux version might be easier

[05:49]

hey JoeBorn

[05:49]

can i have a free beer? :S

[05:49]

<crweb>

by PC do they mean windows?

[05:49]

crweb: it's just a prossibility so far, i think

[05:49]

nerochiaro: I get the impression it's for real, I asked based on may1937 's query and they responded with quick questions

[05:50]

cool

[05:50]

crweb: yes, windows

[05:50]

if possible then, linux would help a lot, but i see that hard

[05:50]

<crweb>

JoeBorn: i would much prefer a linux host.

[05:50]

anyone else?

[05:50]

<crweb>

JoeBorn: if they are talking about giving away the full kit

[05:50]

I think it's just the compiler

[05:51]

<crweb>

JoeBorn: linux first. if you can get both, get both

[05:52]

<crweb>

you'll need to beable to link to the kernel system

[05:52]

<crweb>

and having a linux based compiler might simplify that..

[05:52]

<crweb>

but it all just depends on how the compiler actually works

[05:53]

<crweb>

JoeBorn: did they give you a time frame?

[05:53]

<crweb>

thats fairly exciting..

[05:53]

* crweb is excited

[05:54]

I'm guessing it'll be pretty soon

[05:54]

<crweb>

i've always wanted to write a codec

[05:54]

<crweb>

and with it, we could produce a nice loader interface and work on porting already existing codecs

[05:59]

<crweb>

i think i may have found source code to a C5x mpeg4 decoder/encoder

[05:59]

<crweb>

from franhofer

[05:59]

<crweb>

only, i can't read it to find the actual files

[06:01]

<crweb>

AAC and mp3 for sure

[06:03]

<crweb>

www.iis.fraunhofer.de

[06:03]

<crweb>

the C5x is listed

[06:03]

<crweb>

mm.. lots more reading to do :)

[06:04]

<crweb>

time for sleep though

[06:05]

night!

[06:15]

Joe: I take it you are in China at the mo', or just not sleeping ever again?

[06:26]

sleep, schmeep

[06:26]

* JoeBorn is in China

[06:28]

Yeah, I know what you mean. It will be like that for me again the next time I get involved in a startup or young company.

[07:08]

nerochiaro, ping

[07:12]

meeting, bbl.

[07:52]

turran: in 20 minutes or so, i'm cooking cous cous right now

[07:52]

;)

[08:11]

Hey, nerochiaro, congrats!

[08:11]

Emmett: hey, thanks :)

[08:12]

it's been a long time since i saw you around here

[08:12]

I've been busy! :)

[08:13]

busy of the interesting kind, i hope ;)

[08:13]

Sure thing, back to working in TV and associated tech

[08:14]

glad to head that

[08:16]

Working on a new series right now, hope to have the first block of episodes done this summer

[08:16]

Maybe if I ask Joe real nice, he'll ship the OSD with an episode of it. :)

[08:17]

you gotta make is very short, we are cramped on space on that flash ;) what genre of the show is that ?

[08:17]

www.sonivius.com

[08:18]

science fiction comedy

[08:19]

first thing i thought when i read that page was "hitchhiker's guide"

[08:19]

It's kind of Hitchhiker's/Red Dwarf/MST3K/Doctor Whoish.

[08:20]

and unashamedly so

[08:21]

that's the right spirit. you've got to be proud of these influences

[08:22]

Well, the one thing I hear a bit is 'isn't this kind of like' and they'll mention a show I listed above

[08:22]

and my response is typically, 'of course it is, I love that show.'

[08:25]

so it will air on TV ?

[08:26]

That's kind of an interesting question at the moment, and that decision isn't really mine.

[08:27]

But the powers that be at Sonivius are running all kinds of potential situations to see what would work out best.

[08:27]

good luck then, whatever medium it will end up being distributed on

[08:27]

There's certainly the option of selling it to a syndicate and getting it on TV, and there's already been network interest in carrying it

[08:28]

but on the other hand if we find a way to release it online in a way that pays the bills, that's good, too

[08:29]

The simple question: If you could get a really funny science fiction show delivered to you via the series of tubes, would you buy it?

[08:29]

I would, if it were really good and it were inexpensive.

[08:30]

I might even subscribe to such a thing if it were possible.

[08:32]

well, that's true and i'd probably do it too. however there will always be lots of people who will just share it via p2p, even if it costs very little.

[08:33]

Sure thing, I'm okay with that, too.

[08:34]

The nice thing about keeping it online is that as a company we have the copyright eggs in our basket.

[08:34]

It's not like we'd have to worry about a network pulling episodes offline and suing kids.

[08:34]

which is useless anyway

[08:35]

Right. The idea being floated is that if you pay for the ep, you get bonus footage, blooper reel and extras

[08:35]

if you get the ep on p2p, you wouldn't have access to that stuff.

[08:35]

wait, the extra stuff isn't video too ?

[08:36]

the extra stuff is anything from scripts, SFX notes, music, etc

[08:36]

nerochiaro, hi! congrats for the new job =)

[08:37]

Emmett: right, harder to share live video is. good thinking

[08:37]

some would be video, some would be text, some would be music, you know?

[08:37]

turran: thanks a lot

[08:37]

You're right, no one will ever come up with a way to share music online. :)

[08:38]

The economy of scale would tend to work in our favor. Like a comic book, every episode is pretty much a commercial for the series.

[08:38]

So maybe a million people watch it by grabbing it from p2p

[08:38]

Okay, that's fine

[08:39]

Especially fine as we're not paying for that bandwidth

[08:39]

Emmett: heh, ok. what i wanted to say is that it's harder to share all that material based on a random media than just video.

[08:39]

The people that really get into it, say it's a hundred thousand people

[08:39]

say each of those hundred-thousand are paying us, I don't know, a buck and a half

[08:39]

That's more than enough money to make new episodes.

[08:40]

nerochiaro, ive been into the nms api this days, its kind of unstable, the recorder test app sometimes does record, sometimes it doesn't

[08:40]

and maybe the people that watch it on p2p come to the website and buy a T-shirt, that's nice, too

[08:40]

oh, absolutely. i'm sure p2p sharing will work more like advertising than as a loss.

[08:40]

which is what other bigger players aren't understanding

[08:40]

Plus, you know, I'm not really a copyright absolutist by any stretch

[08:41]

I'd be really, really happy if people like the show enough to want to share it with their friends.

[08:41]

nerochiaro, is there any documentation on the osd devices? (vencode, aencode, tvp5150)?

[08:41]

Emmett: we're on the same page, then

[08:42]

turran: you wish. but no, none

[08:42]

<Ycros>

what show?

[08:42]

turran: at least not open :(

[08:42]

There's a cumulative effect, as well -- If people share it on p2p and millions of people watch it and like it, then that's good for us. It makes networks come to us with gigantic checks to see if they can buy our *next* project.

[08:43]

nerochiaro, :(

[08:43]

Emmett: provided they be persuaded as seeing that as a good thing, yes

[08:43]

turran: however, why do you need the docs to these, specifically ?

[08:44]

<Ycros>

is this a real show we're talking about, or just hypothetics?

[08:44]

Ycros: A real show in production presently

[08:45]

nerochiaro, well im coding an application that should record based on a schedule, i can use the nms api (which is very buggy) or use the devices by myself. but now looks like we have another goal on this project, i need a way to place the current time on the video recorded (osd).

[08:45]

nerochiaro, i can use any gfx lib (i.e evas) to render the text to a buffer, and merge it with the video in

[08:45]

<Ycros>

Emmett: is there a url you can point me at?

[08:46]

nerochiaro, and then encode it to any format. but i don't know the flow of video data, not event the modules api

[08:46]

turran: i doubt you will be able to use the devices without the nms api, i have no idea how that can bedone and the docs, if any, are closed as far as i know.

[08:47]

nerochiaro, yeah i was afraid of that =(

[08:47]

turran: as for timestamping the video, you think of rendering the text on it as it records, of afterwards ?

[08:48]

nerochiaro, better as it records

[08:48]

<Ycros>

that would require doing it before encoding

[08:48]

Ycros: www.sonivius.com

[08:48]

<Ycros>

which happens in the dsp

[08:48]

<Ycros>

which I bet you can't do

[08:49]

indeed

[08:49]

pretty much correct, i am afraid

[08:49]

nerochiaro, btw you might be interested on this.... just added some functionality to ecore_fb (isnt commited yet), to use different keyboard layouts (spanish, french, etc)

[08:49]

nerochiaro, no more hardcoded to english :)

[08:50]

<Ycros>

Emmett: Interesting. Comedy?

[08:50]

yeh

[08:50]

<Ycros>

red dwarf was the first thing that popped into my head

[08:50]

Good!

[08:51]

see above:

[08:51]

turran: oh, good. you know, one other thing i've always wondered about ecore is why it uses strings for keyb events identifiers, it seem a waste to me.

[08:51]

* Ru has quit ("CGI:IRC (EOF)")

[08:51]

Emmett It's kind of Hitchhiker's/Red Dwarf/MST3K/Doctor Whoish.

[08:51]

Emmett and unashamedly so

[08:51]

nerochiaro that's the right spirit. you've got to be proud of these influences

[08:51]

Emmett Well, the one thing I hear a bit is 'isn't this kind of like' and they'll mention a show I listed above

[08:51]

Emmett and my response is typically, 'of course it is, I love that show.'

[08:51]

<Ycros>

hehe

[08:52]

<Ycros>

so, same low-budget awesomeness that red dwarf has?

[08:52]

Same writing staff as the MST3K Pilot That Will Sit On A Shelf Forever

[08:52]

<Ycros>

or is it higher budget ;)

[08:52]

nerochiaro, do you mean the keyname, keysymbol, keycode, etc?

[08:52]

ya

[08:53]

It'll look good.

[08:53]

It'll sound good, too.

[08:53]

it will even smell good ;)

[08:54]

<Ycros>

I haven't seen MST3K

[08:54]

The art staff has told me that their mission is to look better than Battlestar Galactica.

[08:54]

Which I guess is fine.

[08:54]

nerochiaro, its better that way, in the backend you can have whatever keyboard translation you want, at ecore you get the translated code always

[08:54]

<Ycros>

the new one or the old one?

[08:54]

The new one.

[08:54]

<Ycros>

that's impressive then

[08:54]

<Ycros>

because I think the new one looks awesome

[08:54]

All the cameras for Relay 23 are 24-frame progressive DV

[08:54]

turran: why not just numeric constant ?

[08:55]

so it will look a little more 'cinematic.'

[08:55]

nerochiaro, well that's an option too ... but you will have to do all the constants for the utf8 table

[08:56]

turran: i mean, just replacing string constants with numeric constants ?

[08:57]

<Ycros>

Emmett: ETA to pilot? :)

[08:57]

Ycros: This summer. We'll probably end up shooting four episodes at once.

[08:58]

There's CG on both ends of that.

[08:58]

nerochiaro, do you mean the string constants for ecore_fb?

[08:59]

turran: i mean right now the keyname etc are string constants. why not just make them numeric constants ?

[09:00]

<Ycros>

Emmett: don't suppose you guys have a mailing list/newsletter setup so that I know when it's out?

[09:01]

Ycros: When the first episodes are available, I don't think there's a force on this planet that will be able to stop you from hearing my yelling.

[09:01]

<Ycros>

if not, I suggest doing it. Just a little box on the bottom of that page where people can enter their email address

[09:01]

nerochiaro, dunno =)

[09:01]

<Ycros>

hehe, okay

[09:01]

turran: heh, ok, nevermind

[09:01]

<Ycros>

well if you yell in here, I'll hear

[09:02]

With any luck, when it's done I'll convince Joe to ship an episode with the OSD.

[09:03]

<Ycros>

ahh.

[09:03]

<Ycros>

so it'll be distributed for free basically?

[09:04]

nerochiaro, i prefer this way, the keynames is an adobe standard (pdf) and the string representation (keysymbol) is the utf8 value, i do prefer the value in unicode (int) instead of the char * for the value, but dunno if replacing all of this with integer constants is better, as the ut8 is already a constant

[09:04]

Not quite, there's a lot of discussion about that and I'm not really in on it.

[09:04]

But I do have the force necessary to say, 'Hey, let's ship a free ep with the OSD if Joe'll let us.'

[09:04]

turran: well, i'm ok with that too. it was just more efficent with numeric constants

[09:05]

<Ycros>

ahh, okay

[09:05]

<Ycros>

my suggestion would be to release the first ep for free

[09:05]

I think that's inevitable.

[09:05]

<Ycros>

then have the rest downloadable for a small fee per episode, not DRM'd

[09:05]

I would like to do that.