IRC log of #maemo for Thursday, 2019-04-04

sicelo1:02 < DocScrutinizer05> alias n900cam='gst-launch-1.0 udpsrc port=10010 caps=application/x-rtp,clock-rate=90000 ! rtpjitterbuffer !  etc .... what does it do? i'm curious, but don't have a pc at hand to test00:07
DocScrutinizer05sicelo: it starts a video source on N900 from main camera to WLAN, RTP jpeg format/protocol. And same time it starts a viewer on the interactive side (here: my PC)00:17
DocScrutinizer05prerequisite: nothing blocking UDP data on destination port 1001000:18
DocScrutinizer05sicelo: I just fire a `n900cam` in a shell of my PC00:20
DocScrutinizer05(could use a desktop icon or a startmenu entry as well)00:20
sicelohow is the latency? does the n900 cope?00:22
DocScrutinizer05latency ~0.5s00:42
DocScrutinizer05though it's ABR and sort of "hibernates" when long time no change in input scene00:44
*** dafox has quit IRC00:46
DocScrutinizer05when it does (may take up to a minute it seems) then bandwidth of stream goes down from ~2Mb/s to 500kb/s and a "framerate" like 1/s and it takes like 3s to ramp up to narmal operation again. Free motion detection, simply watch bandwidth \o/ ;-D00:49
DocScrutinizer05sicelo: ^^^00:49
DocScrutinizer05works fine00:51
DocScrutinizer05prolly you could reduce latency further by tweaking (or omitting) rtpjitterbiffer00:54
DocScrutinizer05for my purposes it's absplutely fine00:54
DocScrutinizer05absolutely, even00:55
DocScrutinizer05gotta consider how to deal with maemo-camera though00:56
DocScrutinizer05prolly needs some tweaking of dbus stuff, or maybe simply not starting camera during system boot may do the trick00:59
DocScrutinizer05then OTOH I vaguely recall that without camera getting started at least once since boot, the videa had abysmal (or no) white balance done01:00
DocScrutinizer05also on N900 the gstreamer process doesn't get killed on teardown of connection - gotta come up with some smart(er) job management01:16
sicelothanks for the explanation.01:21
DocScrutinizer05yw :-) thanks for your interest01:36
DocScrutinizer05did I share that nice deep-insight "[bash] how to handle signals" stuff?01:54
DocScrutinizer05bottom line: when your script catches a signal that would normally make it abort, you MUST make the script terminate itself by sending that very signal to $self, to let outer shell wrappers know whether to continue or to abort01:56
DocScrutinizer05I.E do NOT use a simple `exit`01:57
*** Venemo has joined #maemo02:02
*** florian has quit IRC02:04
DocScrutinizer05How the shell handles ^C (SIGINT) and why:
DocScrutinizer05TL;DR:       trap 'rm -f $TMP ; trap 2 ; kill -2 $$' 1 2 3 13 1502:24
*** jon_y has quit IRC02:33
*** Kilroo has joined #maemo03:00
*** DocScrutinizer05 has quit IRC03:05
*** DocScrutinizer05 has joined #maemo03:05
*** DocScrutinizer05 has quit IRC03:51
*** DocScrutinizer05 has joined #maemo03:51
*** keithzg has joined #maemo04:05
*** pagurus has quit IRC04:33
*** tm has quit IRC04:43
*** tm has joined #maemo04:47
*** povbot has joined #maemo07:34
*** chainsawbike has quit IRC08:08
*** chainsawbike has joined #maemo08:09
*** Kilroo has quit IRC08:17
*** dafox has joined #maemo09:25
*** hurrian has joined #maemo09:36
*** dafox has quit IRC09:54
luke-jrDocScrutinizer05: if only the boolean result matters, exit 1 works; if you need the exact correct result, then sending signal 2 won't be right for any of the other signals10:04
*** florian_kc is now known as florian10:29
*** jskarvad has joined #maemo11:34
*** eMHa has joined #maemo11:35
*** eMHa_ has joined #maemo11:37
*** eMHa has quit IRC11:37
*** richrboo_ has joined #maemo12:13
*** jon_y has joined #maemo12:20
*** dderby has quit IRC13:21
*** povbot has joined #maemo13:29
*** povbot has joined #maemo13:30
*** Oksana has quit IRC14:50
*** hurrian has quit IRC15:10
*** richrboo_ has quit IRC15:24
*** Kabouik has joined #maemo17:46
*** dderby has joined #maemo18:15
DocScrutinizer05no, the return code does NOT matter, see the article I linked18:25
DocScrutinizer05you may go fancy and set up a separate interrupt handler (trap) for every signal you catch, end send that very signal to $$ (self) after you did clanup18:27
DocScrutinizer05usually this is not necessary though18:27
* DocScrutinizer05 idly wonders if there is, or bash could learn to provide, a $SIGNAL env that contains an array of numeric IDs of signals the process received18:30
brolin_empeyDocScrutinizer05: Somewhat like ${PIPESTATUS} in bash?18:41
*** Pali has joined #maemo18:44
DocScrutinizer05sort of18:46
DocScrutinizer05of course you can do    trap 'cleanup INT' INT; trap 'cleanup ABRT' ABRT; ...18:47
DocScrutinizer05and have cleanup() deal with the signame parameter18:48
DocScrutinizer05or   for (i=1; i<=32; i++); do trap "cleanup $i" -$i; done;  #?18:52
*** dafox has joined #maemo18:57
DocScrutinizer05I'm not sure what the latter will do19:02
DocScrutinizer05not even allowed according to POSIX which only allows names for signal/condition, no integers19:03
*** florian has quit IRC19:38
*** dafox has quit IRC19:39
*** spiiroin has quit IRC20:09
*** florian has joined #maemo20:28
*** florian has quit IRC20:33
*** florian has joined #maemo20:59
*** how900 has quit IRC21:01
*** how900 has joined #maemo21:01
*** how900 has joined #maemo21:01
*** jskarvad has quit IRC21:25
*** pagurus has joined #maemo21:48
*** Oksana has joined #maemo21:50
*** Gizmokid2005 has quit IRC22:26
*** Gizmokid2005 has joined #maemo22:26
*** troulouliou_div2 has joined #maemo22:39
*** spiiroin has joined #maemo22:40
*** troulouliou_div2 has joined #maemo23:05
*** troulouliou_div2 has quit IRC23:27
*** eMHa_ has quit IRC23:44

Generated by 2.15.1 by Marius Gedminas - find it at!