Log from #neuros IRC channel

heya Joe

hi all

Joe: I uploaded some media

hi Improv hows it going?

[15:20]

It goes... *shrug*

[15:20]

I think Shadow12 was looking for you about an hour ago

[15:22]

ok

[16:33]

JoeBorn: you around ?

[16:34]

yes sir!

[16:35]

JoeBorn: i see you are in contact with folks at goole already for the SoC thing ?

[16:36]

not really, I just pinged the program manager over there, since I met him previously

[16:38]

ah, ok, so no application was submitted yet ?

[16:40]

from reading their submission page for mentoring organizations it seems that at least we need to put up a page somewhere with "ideas", that is potential projects that students can pick up

[16:41]

i'm not sure what you think it would be good self-contained projects for folks to do within a 3 months deadline

[16:42]

I dunno, I think we draw from the uncompleted bounties, osd ideas, etc?

[16:43]

ok, i can try to scrounge up a page from these sources which lists potential projects. let me take a look

[16:44]

I've got one: write a N2 song-database parser for the OSD, slap on a N2-esque interface, and voila you can plug your N2 into your sound system through the OSD (the OSD just treats the N2 as an external hdd)

[16:44]

chreekat: one can just pick up the n2 code for the database, even

[16:45]

Yeah, I doubt it would be a 3 month projcet really

[16:45]

JoeBorn: one of the problems i see with getting us into SoC is that we basically require all applicants to have the hardware, or we need to focus on projects that can be completed on PC side only, which are few

[16:46]

or pass em out like candy

[16:46]

haha

[16:46]

heh, silicon candy

[16:49]

I think we can pass them out to anyone that submits an accepted application or whatever

[16:50]

hey our man srobertson is here

[16:50]

presumably back from another drunken rampage

[16:51]

JoeBorn: it's kind of a chicken and egg problem, though. how can students evaluate if they can do the project when they can't access the hardware in the first place ?

[16:51]

well, we can make it clear that we have a no questions asked 30 day return period

ok, didn't think about that

so, surely one of the projects is the package manager, i think

although it might be a bit too little for 3 months project

yep, hopefully our man hellwolf will sign up for that.

[16:53]

well, expand it to include the server side setup, and UI, etc.

[16:53]

I mean one thing that drives me nuts is that there's no central dbase for stuff like yum

[16:54]

I guess it's not supposed to work that way, but

[16:54]

well, it can work like that actually. and probably should

[16:56]

the flickr browser is another thing that might be interesting as SoC project, but also a little too small maybe

[16:57]

I guess what would be nice are open-ended projects with short-term, tangible milestones

[16:57]

chreekat, right and I think both of those fit the bill

[16:58]

I mean sure, it's easy to get some photos downloaded from flicker showing on the TV and declare victory

[16:58]

chreekat: sure, neither of these are strictly defined in terms of design, the student can pull these off as they please

but to do it right with a cool UI that allows browsing down different paths, friends,

searching

displaying when there are new photos from friends, there's a lot of stuff to do

I mean put it this way, how many man hours do you think flickr has invest in their UI?

[17:00]

agree, maybe i should put these things you just said up as suggestions

[17:00]

they have a team of full time people, "running around like roosters" as my daughter likes to say

[17:01]

flickr folks are cool, but they should've not scrapped their old product when they got their big wad of cash from yahoo.

[17:04]

perhaps not

[17:06]

JoeBorn: there was a very fun and loyal community built around that and they wasted it. really a pity. and to think that flickr was born just as "something to keep up afloat"

[17:07]

er "us afloat"

[17:07]

oh, I don't know, didn't know much about that history

[17:07]

<srobertson_>

hello, good people of #neuros

[17:08]

srobertson_: heya

[17:08]

srobertson, was just talking about you

[17:10]

<srobertson_>

"oh, that srobertson, how we miss his channel-flooding neuroticism?"

[17:10]

JoeBorn: i assume instead that the YouTube thing is something we've scheduled to do internally, per your last planning email ?

[17:11]

nerochiaro, I'd think so, but certainly applying the transcoding thing to other formats would certainly be of interest

[17:11]

I was just going on a big rant on this at lunch

well, the rant kind of grew into a bitter argument which ultimately became a fight

for "transcoding thing" you mean the stuff that anders ported for flash video ?

and then a brawl/riot type of thing

a fight ? no knives i hope

[17:13]

but the rule of law is strong in the US, so it didn't grow into any type of revolution or upheaval like that.

[17:13]

yeah, the stuff that anders_ ported but for other things, like wrappers that we don't currently support or codecs like mjpeg

even if they are not real time, I think it would be great to have a "convert overnight" feature

I do that all the time on my PC

but it's a hassle.

and well

<American-Tech>

The cops where paid to keep quiet in the matter. lol

hassle-->rant -->bitter... well you know

[17:15]

and ultimately as American-Tech suggests it ends up being a big expense for neuros :)

[17:17]

JoeBorn: well, i think an "overnight transcoder" thing can be done, but obviously requires people to have at least an USB hard disk and there should also be codecs support for this.

[17:20]

sure, well I need to run

[17:20]

I won't be back until you're presumably fast asleep

[17:20]

JoeBorn: sure. check thiswiki.neurostechnology.com

[17:21]

er wiki.neurostechnology.com

[17:21]

i just jotted down some, feel free to help editing and expanding it

[17:23]

ok, will do.

ok, I'll be off an on here for next few days, at the TI developer conference in Dallas

send me a ????

(email)

have a good one. i'll be on and off too, at the office tying up loose ends. cheers

[17:25]

ok all, goodnight and good luck

[17:25]

bye

[18:49]

hi all, back for a brief period

[22:04]

* crweb is seasoning his new wok

[23:09]

if anyone's alive, how's this for an SoC project -- GStreamer support, at first simply writing output plugins, but later on writing it so that a programmer could use gstreamer instead of nms

[23:09]

no idea what the speed hit would be, but I think the existing work for gst would be nice

[23:10]

<crweb>

srobertson: the speed hit would be huge

[23:10]

<crweb>

we'd have to do all the gstreamer coding in asm to use the dsp

[23:11]

how does the dsp access system ram? *does* it, or is it just a big silly DMA device on a superfast bus?

[23:11]

<crweb>

it doesn't

[23:12]

dsp ram is contained in a subset of the arm's ram

[23:12]

<crweb>

the codecs are kernel modules with addressing outside of the system/kernel allocated ram

<crweb>

srobertson: thats why there is 64mb, but only booting with 16mb

okay, so why couldn't we use a wrapper (icky, but tunable) to the codec interface, much like what's done in the current neuros cooler? (that's an honest question, not a rhetorical one)

<crweb>

srobertson: because thats what nms is..?

well, that interface is closed

touche.

NDA-protected?

<crweb>

srobertson: gstreamer is asmuch codecs as it anything

not even nda, at least i haven't seen it