Log from #neuros IRC channel

<Lewis>

nerochiaro: ok. gracias

Lewis: that's spanish ;)

<Lewis>

nerochiaro: lol

something for the "this is irritating but I don't want to spend 10 minutes figuring out how to enter for the first time" folks

[00:39]

Johndesigner: yes, exactly. something where you just fill in 2 boxes and go

[00:39]

nerochiaro: do you know of any examples of those forms?

[00:40]

i'm actually looking right now

[00:41]

exactly - and that maybe get thrown into a "to be properly categorized" bin

[00:42]

the one at gnome.org is an example of a guided bug submission procedure. it can be simplified a lot, but it's just an example

hrm, what are your thoughts about requiring a registration? If I was just going to report that "such and such" pisses me off...i'd be slightly more irritated at needing to register

you mean bugzilla.gnome.org?

yeah, I just went to the site and slammed some buttons

Johndesigner: users submitting bug reports generally ends in tears. super-generalization, i know, but...

[00:47]

it's nice to discourage the kind of user who would submit without searching at least once for dupes

[00:47]

True, that needs to be easier (or smarter seach) too

[00:47]

Johndesigner: i hear you there

[00:48]

eh heh. I searched for "snort" and got "Zarro Boogs found"

[00:48]

I'm thinking that if we have a general bin to place the quick ones...even the crazy bastards can't mess things up

[00:49]

chreekat: that's a "bug" in all versions of bugzilla

[00:49]

hrm, but then some poor soul needs to go thru all that crap... i haven't thought this through you see

[00:49]

<srobertson> Johndesigner: users submitting bug reports generally ends in tears

heh that funny

and true

i'm only vaguely paying attention here (watching the simpsons), but i can do php if there is something specific we want

sweet

chreekat: write a php script to find all the bugs in the osd and then patch them

;)

srobertson: no problemo, i'll have my cat help by walking on the keyboard

is that how you're supposed to debug? I've been doing it by hand for years, I feel old now.

Basically, we are saying that neuros is this community friendly bunch...that repays the helpful folk by hassling them with a tax form

[00:52]

Johndesigner: without registration you need to have 2 things: a separate are where to stash all crap from unregged users 2) some poor fella that you pay to go thru all that stuff and seaprate the bugs from the spam

[00:53]

Johndesigner: well, i think that drupal 5 will prompt a cascade of web changes; another one of the "it's been six months let's revise everything" events (which I am responsible for at least encouraging if not starting)

[00:55]

interesting...I have to believe that some revamping of bug reporting systems would be in that mix?

[00:55]

Johndesigner: drupal != bugzilla

[00:55]

nerochiaro, Johndesigner: yeah, drupal != bugzilla, but if we're redoing one, that would be a good time to look at the other

sure sure, what was the questions (obvious diversion attempt)

drupal = bugzilla got it

Johndesigner: one thing i feel about bugzilla, is that it should put all these fields that are at the top, like milestone, severity etc, should be in a box at bottom, hidden, showable only with pressing an "advanced" button

or something

Johndesigner: no, the other way around

[00:57]

Johndesigner: "!=" means "different from"

[00:57]

* Johndesigner is a programing no0b

[00:57]

i know, shoul've seen that coming. sorry

and that's a precious perspective to have around

that != was a typo

thanks, nero

* Johndesigner feels better about being a No0b

[00:59]

how do i put the nick in front of message like - "Johndesigner: "!=" means "different from

?

I'm a crappy typer for good measure

tab completion

'j' 'o' 'h' <hit tab>

Johndesigner: depening on your client, either manually or typing the first letters of the neme and then TAB

TasogareNoKagi: ctrl-w on the wrong window?

juuuuuuuust a bit :)

srobertson, nerochiaro thanks

Johndesigner: anytime

I tend to treat any text field like it was vim

* Johndesigner is now twice the man he used to be

mad skillz now

TasogareNoKagi, vim?

though could I get answers to my questions from earlier if possible?

[01:03]

Johndesigner: it's a text editor for unix OSes, and probably others I guess

[01:04]

<nerochiaro> Johndesigner: one thing i feel about bugzilla, is that it should put all these fields that are at the top, like milestone, severity etc, should be in a box at bottom, hidden, showable only with pressing an "advanced" button

[01:04]

agreed, some layout tweaks are needed

[01:07]

Johndesigner: you're welcome to submit a design that you think is an improvement over the current one. We'll scoot some php guru at taking a stab at it

[01:08]

TasogareNoKagi: sorry, can you recap these questions, i can't be arsed to read the log right now

[01:08]

;)

[01:08]

does the OSD have any rw file system storage when it's live, or is it all read only?

there's an rw part, but it's small (and permanent)

nerochiaro, currently the information is like many ppl talking to you all at the same time..hard to dicern where does what ..at least at first.

Johndesigner: i absolutely agree

meaning it's saved through power cycles?

nerochiaro, it on my (rahter long) list of things to cover

TasogareNoKagi: yes, it's a partition of the internal flash

Johndesigner: roger that

nerochiaro: thanks then

Johndesigner: at least no one is ever bored at Neuros ;)

TasogareNoKagi: can i ask you what you want to use that for ?

[01:11]

heh, true, very true. I ahd to go running this am to avoid a panic attack

[01:11]

kidding of course, but much to do, much to do

[01:11]

nerochiaro: I applied for the overnight transcoder idea from SoC, my idea was to base it around cron, which needs somewhere to put the crontabs, at least when they're being edited

[01:14]

Johndesigner: i hear you :)

[01:14]

TasogareNoKagi: there's also non-persistent storage available via tmpfs (a lot like a ramdrive)

speaking of cron

yea that's what I was thinking of using initially, at least until I got things working

i was just having a discussion with Gao on something like that yesterday

oh, btw, back to bug tracking: trac is coming along.

we need a scheduler for stuff (recordings and other stuff)

Maximander: mantis too

I use trac on sites I run with mysql auth off of the site's user tables

works rather well

and i was wondering, is cron the tool for the job or we should put out or own sched daemon ?

Maximander: yeah, i raised trac as an option last "six-month reorg", but at the time it was missing some important features

dependency tracking, blocks, stuff like that

although personally i like trac a lot more than i like mediawiki and bugzilla

yeah, trac is missing alot. but it's coming along, and has very, very active development

i was considering something else than cron because we need to notify application(s) when a scheduled task fires. anyone knows of a good way for cron to do that ?

I filled a bug against 0.10.x on saturday and had a reply in 38 minutes

Maximander: reply or fix ?

nerochiaro: a kludge would be have the scheduled commands be scripts

[01:18]

reply, saying it was fixed in HEAD

[01:18]

nerochiaro: see my project proposal -- ipc could soon be available ;)

[01:19]

srobertson: what kind of IPC you had in mind ?

[01:19]

oh, srobertson: I don't think you were using it anymore, but that server on which you had an accout two years ago is gone

[01:19]

TasogareNoKagi: that's an option, but how you notify apps anyway ?

[01:20]

nerochiaro: you think i actually have thought that far ahead? you give me too much credit. let's just say *efficient* ipc.

[01:20]

srobertson: in case you don't remember, that was meow.bamboozled.org

[01:21]

Maximander: yeah, ftpbot died a long time ago and i never fixed it. heh, that was a loooong time ago

[01:21]

nerochiaro: that's where my knowledge of unix stops :/

[01:22]

nerochiaro: what about DBUS? it's pretty tiny, right? (just throwing it out there)

[01:23]

srobertson: if you still need servers, I have several more now. just not that one.

[01:23]

srobertson: i'm not sure it's that tiny. it could be an option if the library requires veeery little space. what is that based on anyway ? sockets ?

[01:24]

Maximander: i appreciate the offer, as much as i appreciate you giving me a hand then, but i'm good now (Verizon FiOS: 5mbps up)

[01:24]

nice.

[01:25]

nerochiaro: yeah, i'm reading the docs and although it's nice (plus Maemo uses it) it might be a little weighty for 32M exposed RAM

[01:31]

srobertson: but you know what is it based on ?

[01:35]

nerochiaro: whoops. xml over sockets.

[01:35]

SOO FAST

[01:36]

wow, not really what we want

[01:36]

agreed. on the other hand, Knuth was not too far off with "premature optimization = sqrt(evil)" quote

[01:37]

i agree

[01:37]

so, it's almost certainly a no, but i'm not dropping it entirely from consideration

[01:37]

and if the communication is restricted to: 1) app calls scheduler and says "i want to register for notification" 2) scheduler calls app to notify

[01:38]

i think we can go with sockets. maybe not xml over sockets, that seems like a bit of extremely cumbersome for nothing

[01:39]

problem is: apps should keep a socket listening for notifications

[01:40]

or, in the spirit of DCOP, apps could connect to a central (very tiny) daemon which would route requests

[01:41]

why add another daemon ? they can connect directly to scheduler

[01:43]

okay, add scheduled events to daemon. i'd rather think of the daemon as "router" more than "scheduler" because IPC will not be limited to scheduling for long

[01:44]

so you think of 2 deaemons. sched daemon and message router daemon ?

[01:46]

i think it's important to construct a message router daemon first, and poke around with a protocol, before deciding

[01:46]

it wouldn't be hard, though, to add scheduling to the router -- just add a way to "sleep" a message so it doesn't get sent for xxxx (or better probably, until xxxxx)

[01:47]

otoh, this is not persistent

[01:48]

okay, all, i really really have to work now

[01:48]

ok, thanks for the input

[01:48]

nerochiaro: thanks for yours

[01:49]

srobertson: it's just that we're gonna need this stuff to finalize the EPG/Scheduled recording design, so i'm trying to gather all possible idea quickly

[01:49]

srobertson: well, i'll leave you to work and go back to mine. see ya

[01:49]

oh! okay. well, i'm happy to talk more wednesday, if that's not too late --

and of course what i'd "rather" do when coding is subservient to what needs to be done

srobertson: heh, of course

srobertson: well, there's a thread in ML that you can contribute to

srobertson: input on that would be welcome

i shall peruse.

[01:56]

* srobertson is away: HOMEWORK. argh, university, hate crap classes, et al

[03:15]

anders_: have a minute to spare for an xmms2 internals question ?

[03:16]

<anders_>

nerochiaro: Go ahead!

[03:16]

i have just started looking at the way xmms2 does IPC, and specifically the async interface

[03:17]

since we need something of that kind for stuff on the osd, i was wondering how much we could reuse of that system

[03:17]

more specifically

[03:18]

we want to have a scheduling daemon which should notify applications when it's done. and i was thinking that a system like the xmms2 async interface would be good for that

[03:19]

apps register to the sched daemon and the daemon calls them back when done

[03:20]

<anders_>

src/lib/xmmsipc is just lowlevel message passing stuff.

[03:20]

<anders_>

Which has an async interface built on top.

[03:21]

<anders_>

src/lib/xmmsipc should be easy to rip out and reuse.

[03:21]

<anders_>

The higher level stuff is a bit more involved in xmms2 code. And will hopefully be changed to autogenerated code after the summer (SoC project)

[03:22]

<anders_>

Alex recently blogged about the ipc internals; www.geekfire.com

[03:22]

oh, very interesting

[03:23]

i was thinking that since we're going to use xmms2 anyway, this would reduce the overhead

[03:23]

of having both xmms2 ipc and our own ipc

[03:23]

<anders_>

Right.

[03:25]

it uses sockets underneath, for the actual communication, right ?

[03:27]

<anders_>

Yes. Both unix and tcp sockets.

[03:27]

<anders_>

src/lib/xmmsipc/{socket_tcp.c,socket_unix.c}

[03:34]

anders_: the choice of unix or tcp is up to the application ?

[03:36]

<anders_>

nerochiaro: Yes. Application does: xmms_ipc_client_init ("tcp://127.0.0.1:1234/") and server does: xmms_ipc_server_init ("tcp://0.0.0.0:1234")

[03:37]

so you use unix://socket_path for unix sockets or something like that ?

[03:38]

<anders_>

nerochiaro: Right.

[03:41]

anders_: last question, then i'll leave you alone (or maybe go bother someone in #xmms2 ;)) - with a normal installation of xmms2, how much of this stuff is exposed in libraries and that other programs can build against ? the low-level ipc message passing or also the callback mechanism ?

[03:43]

<anders_>

The only thing exposed is xmmsclient.so

[03:43]

<anders_>

Which contains all that stuff inside, but not accessible.

[03:54]

anders_: if we factor that out somehow, can it be of use to your project too ?

[03:55]

<anders_>

nerochiaro: It pretty much already factored out in our project, it just bundles it togheter.

[03:55]

ok, even better

[03:56]

<anders_>

There is a src/lib/xmmsipc/test.c and a src/lib/xmmsipc/testclient.c but I'm not sure they are updated.

[03:56]

<anders_>

And it doesn't seem to use messages

[03:56]

<anders_>

So they are pretty useless.

[03:57]

they write raw bytes to the socket, it seems

[03:58]

<anders_>

Yeah.

[03:59]

xmms_ipc_msg.h is the API for messages, right ?

[04:00]

so it's - create message, - put typed vars into msg, -send/recv msg over socket, - take out typed data from msg ?

[04:08]

good morning :)

[04:09]

<anders_>

nerochiaro: Right

turran: mornin'

hi all

good evening JoeBorn

'lo

howdy

JoeBorn, see you in a few

[09:40]

hi everybody

[09:40]

Johndes|away, see you then!

[09:47]

hello

[10:00]

hi JoeBorn

[10:00]

JoeBorn: seen my email about the TVTV contact ?

[10:02]

yeah, I'm digging out of 135 emails\

[10:02]

JoeBorn: ok, just wanted to make sure, since she's in germany so you might want to contact her before the day expires. it's 5PM there

* vmarks sends email to JoeBorn

* vmarks sends email to JoeBorn

caribbean! who needs tv!

doh.

* vmarks sends email to JoeBorn

* vmarks kids.

h

i

l

o

w

bye all

<anders_>

nerochiaro: Thursday CEST 19:00? Right?

20, we have dst

no wait, what ?

anders_: ?

<anders_>

nerochiaro: Proposed time for GSoC-mentors meeting.

[13:49]

<anders_>

nerochiaro: (I don't have my mail available right now)

[13:50]

anders_: that's correct, but no one answered yet

[13:52]

<anders_>

nerochiaro: I might be able to make it.

[13:53]

anders_: great. hopefully cr-tablet (crweb) will answer too. when you get back to email, please send a reply

[13:54]

<anders_>

I'm not 100% sure though. I'll be out of town.

[13:55]

well, we can reschedule if you prefer

[13:55]

just mention so in the email

[14:00]

well, i'm be off for today. bye everyone

[14:03]

<cr-tablet>

i need a reschedule

[14:04]

<cr-tablet>

I have class at that exact hour

[14:05]

<cr-tablet>

any other hour/time would be good

[17:57]

<pb101>

Anybody familiar with NmsPlay()?

[20:18]

* srobertson is back (gone 18:21:51)

[20:18]

UNDERPANTS

[20:19]

anyone know why I can view some directories through smbclient on the OSD, but the on screen samba browser can't see them?

[20:19]

I've enabled show hidden already

[20:20]

and refreshed

[20:52]

drawnout: forgive me if this is silly, but... permissions?

[21:01]

srobertson: nope, I can view it through smbclient on the neuros

[21:03]

what user does osdmain run as? root, i'm guessing

[21:03]

* srobertson still hasn't pulled out his magic lead wand to fix his serial

[22:03]

<pb101>

anybody awake?

pb101: is that an honest question?

<pb101>

yeah, why not :)

<cr-tablet>

define... awake

* srobertson hugs little bottle of caffeine pills

'scuse me, alertness aids

<pb101>

beging able to respond.

<cr-tablet>

physically or mentally?

<pb101>

mentally, it's more important.

<cr-tablet>

ah..

<cr-tablet>

then no

<pb101>

cr, by the way, I followed your advice, and was able to rebuild the rootfs, boot from NFS, etc...it's all working great!

<cr-tablet>

i know

<cr-tablet>

i designed it

<cr-tablet>

:)

[22:09]

<pb101>

appreciate the help :) I was really approaching the whole thing from a different and way more difficult direction.

[22:10]

<pb101>

I am attempting to add more hooks to the TCP/IP piece, and I am having a problem with NmsPlay(). It's seems to be failing because it can't send commands.

[22:13]

<pb101>

any ideas?

[22:31]

<cr-tablet>

i haven't done any work with nms yet

[22:32]

<cr-tablet>

i've been designing a gui shell, and that will eventually wrap around the nms functions.

[22:32]

<cr-tablet>

i wasn't going to implement nms until the gui was done

[22:34]

cr-tablet: hey, is there a way to disable the console bell for the development VM?

[22:35]

<cr-tablet>

sudo modprobe -r pcspkr

[22:35]

<cr-tablet>

maybe

[22:36]

ah, thank you. loud noise is scary

[22:41]

<pb101>

cool, thanks

[22:43]

<pb101>

drawnout, I feel the same way about the beeping.

[23:27]

is there any way to disable the watchdog timer from the command line?

[23:33]

<pb101>

is it rebooting your box?

[23:34]

it is when I kill osdmain ;)

[23:34]

after a minute or two

[23:34]

<pb101>

there is a bug in the software.

[23:35]

<pb101>

you can get around it by mounting the system using NFS, and simply keeping osdmain from starting by modifying the init scripts.

[23:36]

<pb101>

You can also get the rootfs, modify the init script, build a new upk image, and use it to boot.

[23:36]

*nod* I'll get to that point eventually, but I just got the thing today and have a major research paper due on thursday

[23:36]

<pb101>

that's option a and b. I don't know about another solution...unless you wait until the firmware is fixed.

[23:37]

na, I'm ok with it. I figured out what I need to figure out

[23:37]

<pb101>

it was bugging me too because it kept painting over my app.

[23:38]

well I was hoping to find a way to adjust the resolution since I have a nice lcd tv, but the itfb module is proprietary, so little hope of that

[23:39]

<pb101>

yeah, the OSD is not 100% open-source.

[23:39]

I know, I;m just hoping that's one of the pieces that will be rewritten during SoC

<cr-tablet>

the resolution can be modified

how?

<cr-tablet>

ioctl

ah, it does support it... excellent

<cr-tablet>

however

<cr-tablet>

it wont' do you much good

<cr-tablet>

NTSC doesn't work the same way as pc resolutions

[23:41]

<cr-tablet>

you can make the resolutions as high as you want, its still going to be NTSC

[23:41]

<cr-tablet>

the text you read is just going to get smaller

[23:42]

so no D-1 ntsc?

<cr-tablet>

i don't believe so

<cr-tablet>

its ntsc 720x480

<cr-tablet>

there are actually 2 layers

<cr-tablet>

the gui layer, and the video layer

<cr-tablet>

the "frame buffer" resolution is whatever you want

right now the fb is reporting 656x464

<cr-tablet>

but it doesn't effect output video

<cr-tablet>

the fb is just the gui display

<cr-tablet>

has nothing to do with video out

now is the actual output 656x464 (which is ntsc) or 720x486?

which is D-1

ntsc

<cr-tablet>

656x464 isn't ntsc