#qi-hardware IRC log for Monday, 2013-06-17

qi-bot[commit] Paul Cercueil: Align the links the same way whether or not a scrollbar is present (packages) http://qi-hw.com/p/gmenu2x/4347c3400:56
qi-bot[commit] Paul Cercueil: Remove useless clipper (packages) http://qi-hw.com/p/gmenu2x/a0690ba00:56
pcercueido you guys mind if I switch gmenu2x to use truetype fonts?03:33
pcercueiI want to use the DejaVuSansCondensed font03:35
pcercueiquick preview: http://crapouillou.net/~paul/nobackup/DejaVuSansCondensed2.png03:35
kyakpcercuei: makes sense, i guess05:20
pcercueiwell the main advantage is that we'll be able to display non-ascii characters ...05:21
kyakthat's right.. As i remember, currently gmenu2x is using some native fonts, right?05:24
pcercueithe font is loaded from a PNG05:25
pcercueialso, it'd be about time that we merge the "packages" branch back into the "master" one ...05:44
pcercueiI just need to know that it doesn't break anything on your side05:44
kyakwhat's the "packages" branch for?07:01
pcercueiit makes gmenu2x able to use OPK packages, which is the format we use in OpenDingux07:03
pcercueiif libopk is found, support for OPKs is enabled, otherwise it's disabled07:03
pcercuei(at compile time)07:03
kyakwhat does it mean - use opk packages?08:07
kyaklike, being able to install them?08:07
mthkyak: OPK is based on squashfs, so packages are mounted when you launch an application, they are not installed14:03
viricwhat is opk?14:05
viricthe openwrt packager?14:05
kyakmth_: i guess it doesn't apply to openwrt then?16:56
mthviric: OPK is the application / package format we use for OpenDingux on the GCW Zero (and on the Dingoo A320 in the future)18:04
mthkyak: you could build OPK files for use on openwrt, as long as you have squashfs in the kernel18:05
mthit's a different approach than package management though18:05
mthwe've got a lot of preinstalled libraries in the rootfs18:05
mthand no dependency mechanism18:06
mthso every lib you neeed that is not in the rootfs has to be added inside the application package18:06
kyakmth: ok, i see. I thought you were talking about openwrt's opk18:13
viricgcw zero? mh18:13
mthisn't that opkg?18:14
mthviric: http://www.gcw-zero.com/18:15
kyakmth: well, the file name end with opk, hence the confusion :)18:20
viricah nice thing18:21
viricI wish I knew all the games there in screenshots18:22
mththe "media" tab has the game names in it18:23
viricaha! thank you18:24
viricmth: can this thing play ogg or mp3 with low power?18:25
viricI wondered, what open device allows to play mp3 (with loud speakers) without taking much poewr18:25
viricthe kickstarter looks like it worked! are you earning money with this project?18:26
viriclucky those, who do :)18:27
mthyes, I am being paid, although it's not comparable to a normal software engineer's salary18:29
mthbut it's fun work, all open source and I learn a lot, so for now that is not a problem18:29
viricwell, I imagine so.18:29
viricBut It looks fun18:29
mthI got about 6 hours of emulator use out of one battery, for music playing the LCD + backlight can be turned off and the CPU usage would be much lower, so it would last considerably longer18:30
mthI don't know how much though; we haven't measured the power use of individual subsystems yet18:30
mthpcercuei did that for the Dingoo A320, but he doesn't have access to his equipment currently due to being on the other side of the world18:31
viricI mean, music is decoded with the general cpu, right?18:31
viricsomehow I imagine it far less efficient than with hw help18:31
mthfor today's CPUs music decoding is so easy that they'll sleep most of the time18:32
virichm really?18:33
viricwell, the gcw also has a huge battery :)18:33
mthplaying MP3 takes about 6% CPU time18:36
virichm ok18:37
viricI should test how long the nanonote lasts :)18:37
viricthe battery is far smaller18:37
mththe JZ4770 is manufactured at a smaller process size (65 nm vs 180 nm for JZ4740, if the wikipedia page is correct), which also helps power efficiency18:39
viricah, ok18:40
mthwe can probably save some power in the kernel as well18:41
mthwe don't unpower all unused hardware yet18:41
viricvery good :)18:42
viricyesterday I was trying 3.9 in a broadcom chip... I'll try to build a filesystem for it.18:42
viricmth: do you have dropbear there?18:43
viriccan you try with dbclient:    ssh -J hello hello18:43
viricif it segfaults or not18:43
mthah, the ssh client from dropbear18:44
mthI read db and think of databases ;)18:44
viricthe dropbear ssh client18:44
viricit's about bears18:44
mthopendingux:/boot/local/home # dbclient -J hello hello18:44
mthSegmentation fault18:44
mthopendingux:/boot/local/home # sh: hello: not found18:45
viricalso segfault? Damn db18:45
viricon x86 it works.18:45
mthI only use the server side18:45
mththe Zero doesn't have a keyboard, so the client is not that useful18:45
viricthis requires gdb!18:45
mthI've got gdb18:45
mthgdb needs a patch to work on MIPS though18:46
viricyou could have an autossh trying to forward ports :)18:46
viricReally? I used gdb on mips enough times18:46
viricwhat patch?18:46
viricsince when?18:47
mthhere is the detailed info: http://sourceware.org/bugzilla/show_bug.cgi?id=14523#c318:48
virichm I didn't remember it. Thank you!18:49
mthif you're using 7.3 or before, you don't need this patch18:50
viricAh, that can explain it18:50
viricThis channel is a great place to be at :)18:50
mththe segfault happens on free(), by the way18:51
viricI'll try with valgrind on x8518:51
mthbut I don't have debug symbols in dbclient, so gdb cannot provide a good stack trace of the upper levels18:52
viricsome hundreds of problems. perfect.18:52
viricdon't worry18:52
mthit could also be a corrupted stack though18:52
viricif so, valgrind will tell on x8618:52
viricwill check. don't worry. :)18:52
viricbut this bug has fcked me on openwrt routers18:52
viricand it's there since years18:53
viricmth: can you test a patch on dropbear easily, for mips?19:20
--- Tue Jun 18 201300:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!