Log from #neuros IRC channel

well i've tracked my problem down to nmsrecord. But this is still only a wrapper which send the command BOOL sendret = sendNmsCmd(&rdata, CMD_RECORD, data, dlen);. But who receives this command and where would the code that does CMD_RECORD be implemented. Because i'm really having some difficulties making the Neuros believe that it writes to a real disk

[00:57]

mgschwa1: you should speek to nerochiaro then when he comes online in 1 hour or so

[01:04]

ah one hour could be too late for me, i have to my course. Do you know how long he is going to be here. Because i really need to know what the Recording on neuros does with the filesystem prior to writing the actual data. Otherwise my streaming won't work

[01:30]

<crweb>

he willl ususally be here from 07:00 UTC to about 18:00

[02:28]

mgschwan: you still there ?

[02:48]

hi nerochiao, sorry my internet connecting is somewhat unstable. It seems that the problem is not my filesystem but that asf files can't be recorded on my osd, because it does not work with my nfs share either. With mp4 files it starts broadcasting them over my lan, but that does not help because i can't view them

[02:52]

nerochiaro, are you still there?

[02:53]

mgschwa1: yes

[02:53]

mgschwa1: does asf recording work without your modifications ?

[02:53]

are there any special requirements to record asf files?

[02:54]

mgschwa1: i don't think there are. main issue we found with them is that appearently you can't record over 2Gb

[02:55]

mgschwa1: but there's also a weird issue at the start of recording we are investigating, that involve some bad seek into the file, might affect your fuse fix

[02:55]

mgschwa1: how do you handle seeking anyway ?

[02:55]

i tried saving to the nfs share (which works with mp4) but i get the error message "WARNING: ./../src/nw-video-recorder.c--VideoRecorderStart: 1133

[02:55]

Unable to record. Please report this server error code as a bug: 8-42:-1.

[02:55]

"

[02:55]

with or without your changes ?

[02:58]

i did no changes on the nfs filesystem. I wrote a fuse filesystem and compiled it with arm-linux-gcc and simply copied it to the OSD. That means i even didn't compile my own image (well i did but i didn't install it on the osd). The firmware is 3.33-1.73-00.82j

[02:58]

sorry, not 00.82j but 04.81t

[03:00]

i currently don't implement seeking (just a stub) but it starts sending the mp4 data over the network. Maybe i'll go and try avi

[03:00]

i think we record only mp4 and asf currently

[03:00]

and i'm pretty sure at some point the asf encoder tries to seek

[03:01]

but in any case you get that error with NFS, so your filesystem isn't used, right ?

[03:01]

yeah i was guessing that seeking is necessary but first i wanna get the recorder to do something before i start dealing with the next step.

[03:02]

ok. can you record ASF to a media card or USB disk, after a reset ?

[03:02]

Thats correct my filesystem does nothing it is mounted to a different directory

[03:02]

wait a sec, i try

[03:02]

thanks

[03:03]

i'm going to remove my fs from boot script, that i can be sure that nothing unexpected comes into my way.

[03:04]

good idea

[03:12]

mysterios, well i think i'm trying an upgrade

[03:12]

too bad my system is already up to date (at least in the official tree)

[03:13]

mgschwa1: mysterios ? what do you mean ?

[03:13]

i sent you something in private. Didn't you get it?

[03:14]

no, i think you need to be registered with freenode nickserv to send me PM

[03:14]

to send anyone PM

[03:15]

and now?

[03:15]

should i switch from Official to [Beta] ?

[03:16]

mgschwan: don't paste code in that way you're doing in my PM. use pastebin.ca or similar pastebin site

[03:16]

code pasted in IRC is hard to read due to timestamp, line wrappig etc

thanks

reading

you don't need to thank me, it was my fault

[03:19]

what i see there is in one case the media server crashing, in the other not being able to record due to fault in the asf plugin. this is really strange, since both recording ways should work ok, at least to MMC card

[03:19]

if what i'm trying eventually works (if ever). I am going to use the file name as the Broadcast address (i.e. 224_0_0_1-10000.asf) would send the stream to 224.0.0.1:10000

[03:20]

that would be quite cool

[03:20]

yeah it crashes because i hit Ctrl-C a few times, because i didn't really wanted to record just to test it :)

[03:21]

hmm, wait a second. are you recording using the normal OSD GUI application or using some test app you made ?

[03:21]

The other way around would be cool to. Mapping a Broadcast stream into the filesystem. That's by the way the reason i started it in the first place, but i have nothing working now

[03:22]

i mean, to record you press RECORD key ?

[03:22]

i am recording by typing recorder in the telnet sessiong because i wanted to see the output (because i don't have my serial cable with me=

[03:22]

but i'm going to try it throug the gui now

[03:24]

don't think it will change much

[03:24]

but just to be sure

[03:24]

well that's interesting. the recording through the menu worked ?!?!

[03:24]

I'm now going to try if it works with my filesystem too

[05:27]

<crweb>

JohnDesigner: to

[05:27]

<crweb>

yo even

[05:28]

hello

[05:28]

<crweb>

about our.. font

[05:28]

<crweb>

I have to make sure it contains all latin characters

[05:28]

<crweb>

for euro, english, etc

[05:28]

ok

[05:30]

<crweb>

i'd assume being the greatness of a graphic designer that you are. Fonts would be more your thing :)

[05:30]

<crweb>

Don't let us down

[05:32]

* crweb sits tapping fingers on desk waiting for JohnDesigner

[05:32]

crweb: sorry, still not sure what you are asking me

[05:32]

<crweb>

i'm not sure either :)

[05:32]

hehe

[05:32]

<crweb>

the .ttf we have

[05:32]

<crweb>

it supports all Latin langugages or just english?

[05:33]

as delivered, that .ttf contains all the characters it is going to contain. I can't say for sure if that includes all the needed latin characters as I don't know what those are exactly.

[05:35]

<crweb>

well, we are about to ship full language support..

[05:35]

<crweb>

needs to be 1 font

[05:35]

<crweb>

all latin characters i mean

[05:35]

<crweb>

so, if we don't have this, we need new font

[05:35]

how much of the font table did you delete to make it smaller when you OSD-ized the font a while ago?

[05:36]

* crweb shrugs...

[05:36]

<crweb>

i guess thats my question, did the .ttf have a full font table..

[05:36]

let me finish something here and I'll see if I can help you somehow. gimme 15 mins or so.

[05:37]

<crweb>

ok

[05:38]

ya, that .ttf has the full font table AFAIK. Actually, I'm not sure how to define a "full dont table" to be honest.

[05:38]

<crweb>

i heard users complain that TORFU font wasnt complete..

[05:38]

<crweb>

greyback would have more info on this

[05:39]

well, if that complaint is what I am thinking it was, that issue may not have been font related after all, can't remember really.

[05:39]

<crweb>

square boxes showing up in filebrowser instead of characters

[05:39]

ya, i kinda sorta remember that.

[05:40]

Did you delete part of the font table to make the font size small enough for the OSD?

[05:40]

<crweb>

i deleted what I didn't think was useful

[05:40]

<crweb>

i'm pulling it up now

[05:41]

how did you determine what was and wasn't useful?

[05:42]

<crweb>

i unchecked everything but Basic Latin (0000 - 007F)

[05:43]

<crweb>

there are about 300 options though

[05:43]

<crweb>

everything from shapes to Georgian.. and many marked Extended or Supplement

[05:43]

I see

[05:44]

<DeepB>

� � �-� and �-�?

[05:44]

its a no go for size if including the whole font table (just to be sure)?

[05:44]

<crweb>

its a little excessive yes

[05:45]

<crweb>

expecially since if the ttf doesn't have those characters, they still take up space

[05:45]

<DeepB>

and � �?

[05:45]

<crweb>

DeepB: people shouldn't be allowed to use characters I can't type lol :D

[05:46]

�h ??

[05:46]

<crweb>

JohnDesigner: i'll check sizes just to be sure.

[05:47]

??��

[05:47]

d�� nd�r ��nt�d st�ff !

[05:47]

<crweb>

JohnDesigner: yeah, try with the full table font size is 9x larger

[05:47]

<crweb>

nerochiaro: i hope that was worth it

[05:48]

crweb: it took 3 seconds

[05:48]

nerochiaro: you win ;)

[05:48]

i have an .it keyboard, remember

[05:48]

<crweb>

foriegn freeks..

[05:48]

<DeepB>

crweb: burn the poor people like me!

[05:48]

<crweb>

lol

[05:48]

crweb: so is 9x larger too large or ?

[05:49]

<crweb>

JohnDesigner: oh thats 9x larger for each size

[05:49]

<crweb>

yes, way to big

[05:49]

<crweb>

was 32kbytes, after is 288k

[05:49]

crweb: I have every confidence that you will do the right thing and make us all proud!

[05:50]

<crweb>

well, i don't want to include a bunch of codepoints (characters) that don't actually have data in them cause the .ttf didn't fill them in

[05:51]

<crweb>

we might need mr foreign keyboard showoff to display some files on the osd and see how it does though

[05:51]

<crweb>

nerochiaro: if you could please :)

[05:51]

what do you need ?

[05:51]

<crweb>

name some files with a bunch of strange characters

[05:51]

<crweb>

see if they show up

[05:52]

ok

[05:52]

Sounds like you'll need to delete individual characters rather than categories (unfortunatly)

[05:52]

<crweb>

JohnDesigner: well, we don't need Arabic, Hebrew, Chinese, etc sections

[05:52]

<crweb>

or Canadian

[05:53]

<crweb>

and Runic ?

[05:53]

crweb: sounds like you have a good handle on this

[05:54]

<crweb>

except that i don' know how to see characters, or how fonts actually work ;)

[05:54]

it doesn't just show the accented chars

[05:54]

i tried it

[05:54]

<crweb>

alright. let me enable some more stuff

[05:55]

"fo�" shows just "fo"

[05:55]

<crweb>

do you know of those are Latin-1 Supplement, or Latin Extended-A etc?

[05:55]

i have no idea

[05:55]

i think some of these are in base ASCII

[05:55]

possibly all

[05:55]

but i may be wrong

[05:56]

guys i've created a new screen and i am launching it from main-menu.... now from my new screen i want to re-raise the main-menu again... how can i do that

[05:56]

BIMZIE: you need to close your app

[05:56]

<crweb>

BIMZIE: i suggest having Qt::Key_Escape trigger your app to close

[05:57]

BIMZIE: or use NAppChannel to ask nwm to run main-menu again, maybe

[05:57]

nerochairo: when i close it... this will re-raise the main one....

[05:57]

<crweb>

BIMZIE: yes, nerochiaro wont work

[05:57]

ah, ok

[05:57]

<crweb>

2nd app request is dropped of app is found running

[05:58]

<crweb>

i figure we'll get more advanced in this area when the time comes

[05:59]

ok

[06:00]

but when i kill my new screen process from konsole... it creates an new main-menu process

[06:00]

and none of them seem to work

[06:09]

<erwan_taf>

hey folks

[06:10]

<erwan_taf>

I saw on the gsoc that you plan to add some DLNA sutff in your products ?

[06:10]

erwan_taf: we already had some, but were forced to disable it

[06:10]

but yes

[06:10]

<erwan_taf>

nerochiaro: \o/

[06:10]

<erwan_taf>

I was on that chan a long time ago

[06:10]

<erwan_taf>

I'm owning a neuros ;)

[06:11]

i'm not sure how far that will go. some people are having major problems with upnp stuff

[06:11]

because of crappy hardware

[06:11]

<erwan_taf>

I'm also participating in coherence

[06:11]

so it's hard to explain to people why sometimes it work, and somtiems it doesn't

[06:11]

so that's why we disabled it currently

[06:11]

erwan_taf: i remember you ;)

[06:12]

<erwan_taf>

I saw you were speaking about making some neuros being compliant with mythtv

[06:12]

erwan_taf: yes, that means using upnp av

[06:12]

i think

[06:12]

<erwan_taf>

coherence is also serving files using upnp and so more that just av ;)

[06:13]

<erwan_taf>

it is also picture/music capable

[06:13]

<erwan_taf>

I did connect my ampache db to coherence and it works

[06:13]

av includes media server, which is serving files

[06:14]

<erwan_taf>

you mean you want to serve file from your product or to watch upnp content on your prodcut ?

[06:15]

the second thign you said

[06:15]

watch content

[06:15]

that is served by other devices

[06:16]

<erwan_taf>

ok

[06:17]

<erwan_taf>

I was searching if any syngergy between coherence & neuros could append

[06:17]

<erwan_taf>

does the OSD beeing developped in C ?

[06:19]

c++ now, but we also use c stuff when needed

[06:20]

<erwan_taf>

ok

[06:21]

<erwan_taf>

thx for the refresh nerochiaro

[06:21]

<erwan_taf>

I haven't been here for ages !

[06:21]

you're welcome

[06:22]

* erwan_taf still have its old neuros in his car :p

[06:22]

<erwan_taf>

the logo is very nice

[06:22]

<erwan_taf>

(duck)

[06:24]

it was made by one community member

[06:29]

<erwan_taf>

how does the mentoring ?

[06:29]

<erwan_taf>

did you had applications ?

[06:30]

end of week we are going to rank them

[06:30]

* erwan_taf is mentoring on syslinux ;))

[06:31]

:) hey fellow

[06:42]

<crweb>

how is syslinux doing these days?

[06:45]

<erwan_taf>

pretty fine

[06:45]

<erwan_taf>

hpa is currently integrating gpxe to extend its capabilities

[06:48]

<crweb>

neat

[07:06]

<erwan_taf>

yeah really

[07:06]

<erwan_taf>

we also plan to directly integrate gfxboot as a com32 module

[07:30]

crweb and nerochiaro guys i've done that by key_escape way.... but now i want to ask u guys a thing if i want to tab focus... do i need to do everything from scratch or there is some routine or any api for that.....

[07:34]

nerochiaro, are you still there?

[07:35]

kind of

[07:35]

mgschwan: on and off

[07:36]

mgschwan: did it work using the GUI app ?

[07:36]

:) sorry, my connection dropped and it was time for me to leave for the university

[07:36]

BIMZIE: tab focus inside the same application ? between widgets of the same applications ?

[07:36]

mgschwan: no problem

[07:37]

mgschwan: are you still having that problem ?

[07:38]

yes it worked with the GUI (which ic quite interesting). It even worked with my Filesystem. It broadcasted the recording to my Local LAN, but as expected neither vlc nor mplayer could play it (probably du to some seeking of the recorder). But at least mplayer told me that he found some video data. And by saving the whole stream to disk it even looks like asf, at least "file" tells me that it is Microsoft ASF

[07:39]

hello?

[07:39]

i would need to know where the code that actually stores the data on disk is located in the source tree so that i can try to find a solution for the seeking problem. My first guess is a simple buffer that allows the recorder to seek in a given Window. Depending on how far the writing goes back this could work.

[07:40]

mgschwan: it would help me if in your file system you could actually trace if there is in fact any seek done by the recorder

[07:40]

nerochairo: yes jus like we use tab key in desktop applications

[07:40]

do i need to setup all the keys action myself

[07:40]

ahhh, thats what i was looking for, thanks

[07:40]

mgschwan: unfortunately that code is part of the closed source section of the code :(

[07:40]

oinhno: hi

[07:40]

hi

[07:40]

like on arrow down..... set focus to next widget

[07:41]

i don't really understand half of what you are sayong

[07:41]

saying*

[07:41]

oinhno: because i'm not saying that to you ;)

[07:41]

i no, i just dont understand that computer stuff

[07:41]

so nerochairo is there any thing like that in it.... or should i do it myself....

[07:42]

BIMZIE: look at how the other apps do it, i actually don't remember, but i think it's handled automatically

[07:42]

i only know the simple stuf

[07:42]

oinhno: what are you trying to do ?

[07:42]

trying to find a suitable irc chat channel to my needs

[07:42]

ok no problem. I don't really need to see the code i just would need to know how the asf file is written (what type information it is that has to be written after the data has been written, i guess that would be the packet headers). The fuse filesystem don't have a hook for the seek command, instead the write function has an offset parameter (that means the seeking ....

[07:42]

oinhno: what are your needs ?

[07:42]

is done completely by the librrary or by the VFS

[07:43]

but i still can check if the offset are monotone rising

[07:43]

rs and other things, nothing important

[07:44]

Currently i am still at the university which means i can't test anything.

[07:44]

... i'm only in high school

[07:45]

bye then

[07:45]

mgschwan: how it works internally, is in two levels. one is the ASF library from vendor, which is closed source even to us. this library use some callbacks to write to disk, which we implement ourselves (basically it has a write and a seek callback, which we implement with some buffers, but in fact it's more or less like if we are calling the fs functions directly from these callbacks

[07:45]

)

[07:45]

weird guy

[07:46]

mgschwan: so all this means that i'm not actually sure myself how much the ASF encoder seeks or not

[07:46]

well do you allow seeking outside your buffer (that means: does this buffer only act as a cache, and if the cache is missed you access the fs directly)

[07:47]

mgschwan: yes

[07:47]

mgschwan: i got to check but i think yes

[07:47]

* nerochiaro looks

[07:47]

Well i simply need to output the write offsets that will show me if and how far they seek, but as i said not before evening

[07:48]

mgschwan: if you do, that would help a bit. i'll actually try to trace something similar here by putting traces in the seek callback function

[07:48]

mgschwan: where are you ? europe ?

[07:49]

The guys in the forum who are using the "write to network share and play the file while it's still written" play the file and if the asf encoder would write something at the beginning of the file while they are in the middle, their player would not recognise it either so i guess that a buffer with enough space for seek backs would work

[07:50]

yes, austria

[07:50]

mgschwan: that may be, i honestly started looking at asf encoder myself not that far ago (what little we have of it, actually)

[07:51]

mgschwan: i'll be around i think this evening, your time

[07:51]

(i'm in the same timezone as you, italy)

[07:52]

:) perfect, but i could remove the udp broadcasting code from my source and send you the file (currently it's only one .c file ) if you want to produce the logs yourself

[07:53]

mgschwan: i think i'll use the code i have and make them in the callback, just to have two different ways

[07:53]

mgschwan: not in a hurry

[07:55]

<crweb>

i think its because of how we write teh files

[07:55]

which is ?

[07:56]

<crweb>

with all that finalizing and scratch area stuff

[07:56]

i am also thinking the other way around. If the broadcasting works it would also be cool to receive a broadcast (or unicast) and map it into the filesystem, which is the reason i started it in the first place

[07:57]

crweb: that's mp4, not asf

[07:57]

mgschwan: agree, would be nice

[07:57]

<crweb>

ah

[07:57]

crweb, but they forums.neurostechnology.com are playing the incomplete file that means there must be some way to do it

[07:57]

mgschwan: makes sense to think so

[07:57]

The mp4 saving seems quite strange, at least for an outsider like me

[07:58]

mgschwan: mp4 is ugly. it need to keep in memory a cart load of stuff, and finally write it on the bottom of the file, then go back to update the head of the file, IIRC. but we can't keep in memory that stuff, so we write it to some temp files

[07:58]

are there other formats supported by the library you are using

[07:58]

* mgschwan is thinking about things like TS, MPEG2

[07:59]

mgschwan: no, unfortunately right now only asf and mp4

[08:04]

do you have any memory structures in you libraries like a queue or ring-buffer. I want to make a buffer (i.e 128Kbyte) that is filled with the data from the write calls at the corresponding position and when it's full it starts sending the data over the network, this would allow the asf writer to seek back to the data in the queue and leave a short timeframe for the seeks, all write operations outside the buffer will simply be ignored

[08:05]

nothing really clean that can be re-used easily

[08:05]

ok then i'll going to write it on my own

[08:06]

or steal it via google :)

[08:06]

there's some very nice ringbuffer code inside the xmms2 source code that can be probably be ripped off fairly easily and re-used

[08:09]

ok i'll go and look

[08:12]

mgschwan: i'm reading some more the asf source

[08:12]

mgschwan: and it say that in the begginning it's setting the ASF header in "broadcast" mode

[08:12]

which probably means streaming

[08:12]

and at the end it seek back and write a non-streaming header

[08:13]

or something like that. i don't have the full source as i said before

[08:17]

mgschwan: it seeks only 2 times, and both times it's seeking to position zero: once when i start the recording, and then once when i stop the recording. both times it's probably calling the same internal function to write the header

[08:17]

mgschwan: once in "broadcast" mode, the other with real lenght

[08:19]

when does it seek to the position zero, after some data has already been written or has no data been written before the seek?

[08:20]

well i have to look into this

[08:22]

<nick_fn>

how do people set the clock on their TORFU OSD? I see no ntpd, or ntpdate

[08:23]

i'll stop my coding on the buffer version until i can verify that the file can't really be played, maybe i lost some packets and thus could not play the file

[08:23]

nick_fn: no way to auto update now, you just set it as usual

[08:23]

manually

[08:23]

<nick_fn>

after every reboot?

[08:24]

no, once

[08:26]

<nick_fn>

pretty sure my dhcpd gives out NTP details. I guess I file a bug/request...

[08:26]

nick_fn: search bugzilla, there may be one or more already

[08:27]

<nick_fn>

I tried out lpkg just now. It worked (it installed stuff.)

[08:28]

nick_fn: it's MattJ's good work :)

[08:30]

<nick_fn>

I voted: bugzilla.neurostechnology.com

[08:32]

thanks

[08:32]

<nick_fn>

only 3 votes :-/

[08:32]

Well i gave it a try and used ffplay on the stream i saved earlier today (only about 1M). It can't play it but the output looks quite promising. [asf @ 0xb7f742d0]Could not find codec parameters (Video: mpeg4, 352x288)

[08:32]

and mplayer tells me that there is a audio and a video stream

[08:33]

<nick_fn>

wonder if the OSD will play back the TS stream I get from my DVB-T card

[08:33]

according to neochiaro, no

[08:33]

<.. nick_fn

[08:33]

<nick_fn>

mgschwan: oh

[08:33]

mgschwan: i was only saying that for recording, not playback

[08:34]

<nick_fn>

I'll test it for playback.

[08:34]

<nick_fn>

I find recording on the OSD not useful for me. The results never play on any of my kit...

[08:34]

nick_fn: thanks for vote on that bug, i hope if will be put in schedule sometime soon

[08:34]

well that opens a whole new bunch of possibilites

[08:35]

nick_fn: thankfully for many others recording works well enough ;)

[08:36]

<nick_fn>

nerochiaro: I tried to record things for my Nokia N800, but it just doesn't play nicely, I have to use my desktop to rencode after.

[08:37]

nick_fn: maybe you choose the wrong recording params

[08:44]

neochiaro i still have one question from earlier. The first seek to position 0. Is this done after some data has already been written or is it really the first thing that is done (which would be quite useless, seeking to position 0 in a file that is empty :) )

[08:46]

mgschwan: i am not sure, but i think it's the very first thing. i think that it just call something like writeheader function, which internally seek to 0. of course the seek to zero makes sense only at the second call (which is at end of rec), but the first one does it anyway

[08:46]

i don't have the source of that function that i called writeheader so i don't know for sure

[08:46]

but it looks like that

[08:47]

ok i'll look into it

[08:47]

thanks, i have to leave for my course now

[08:47]

have a nice lesson

[08:48]

thanks

[12:12]

hi all, hows everyone doing?

[12:16]

I saw Meshuggah (swedish metal) last night, and drank more than I intended. I apparently discovered our kitchen ceiling fan ("Holy crap! When did that get there?!"), which my roommate thought was hilarious

[12:16]

so my neck is a little sore and i think a hangover is lurking

[12:17]

how are you doing? :)

[12:29]

well, I'm doing fine :)

[12:29]

afaik, there are no new fans around here.

[12:30]

well that's good

[12:57]

<gremlin[it]>