[maemo-developers] Problem with Home plugin
Hello All, Could you please help me with the issue I have with Maemo? When I update my omweather home plugin package https://garage.maemo.org/projects/omweather/ on Nokia 770 I have to reboot the device because plugin isn't unloaded from the memory. Switching plugin off doesn't help. The same thing happens with any home plugin. The lsof output clearly shows that plugin's stay in memory even when they're switched off. Here is example output of lsof when plugin iphome is switched off: Nokia770-39:~# lsof|grep iphome > > maemo-lau 856 user memREG 31,4 11692 40758 > > /usr/lib/hildon-home/libiphome.so > > Nokia770-39:~# > In this tutorial http://maemo.org/platform/docs/howtos/maemo-plugin-tutorial.html#home-plugins I found that hildon_home_applet_lib_deinitialize should do the job: > > /** > > * @hildon_home_applet_lib_deinitialize > > * > > * @param applet_data Applet data as returned by applet_initialize. > > * > > * Called when Home unloads the applet from memory. > > * Applet should deallocate all the resources needed. > > * > > */ > > > > void hildon_home_applet_lib_deinitialize(void *applet_data); > > > > This very function when switching off the plugin is not called. Is this > > a bug? May be I miss understand the problem? > > > However it seems that this function isn't called when switching off plugins or just doesn't unload plugins from memory. Is this a bug or have I missed something? How should it be done properly? Thank you. Vlad Vasiliev ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: [maemo-developers] Unreliable Large Network Transfers
On Fri, 2006-12-15 at 21:31 -0500, ext Mike Lococo wrote: > Hi Folks, > > >> I'm finding that large (>200MB) transfers of data via the wifi network > >> are extremely unreliable. > > > > Are you sure network is a problem? Are you writing data somewhere? Both > > MMC card and iternal flash are really slow when writing. > > This is the issue. I can reproduce both the slow write speed (~150KB/s) > and the GUI unresponsiveness with (from memory, please forgive any typos): > > dd if=/dev/zero of=/media/mmc1/test.file bs=1M count=250 I asked around and there seems to be a known kernel problem with this, but there's a workaround: it should work if you use 4k block size. BR, Kimmo > > which doesn't involve the network at all. > > I may do some playing with multiblock write kernels at this point, > although it's not so much the (average) speed that's the problem. I > think the real issue is that the CPU doesn't seem to be able to service > requests from other subsystems in a timely manner once the write cache > fills up. Userspace programs are getting regularly starved for CPU time > for several seconds at a time, which causes the wildly fluctuating > transfer rates, the unresponsive GUI, the Unison sessions dropping, and > possibly the reboots (as programs aren't responding to the watchdog > promptly?). > > The Unison failures, in particular, are disappointing. It is such an > awesome program for syncing the 770 to a desktop system, but if you have > a large MMC (and you change its contents regularly) it isn't usable at all. > > Thanks, > Mike > ___ > maemo-developers mailing list > maemo-developers@maemo.org > https://maemo.org/mailman/listinfo/maemo-developers ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: [maemo-developers] root image scripts appear to be unavailable
On Sun, Dec 17, 2006 at 01:08:07PM +0200, Ferenc Szekely wrote: > Hello, > > ext Aaron Levinson wrote: > > The root image scripts that are referenced by the "Using the flasher > > utility and creating the referencfe root file system" how-to no longer > > appear to be available. According to this Web page, they can be retrieved > > at https://stage.maemo.org/svn/integration/trunk/rootimagescripts/ , but > > this link does not work at the moment. I also looked for the scripts in > > the maemo SVN tree but didn't locate them there as well. > > > Thanks for the report. The underlaying Berkeley DB had to be > "recovered". The scripts are available again. Uh, oh. I don't trust Berkeley DB. It tried to eat my Subversion repository that I used for storing my home directory backups. Not to mention the frequent wedges requiring admin intervention. I strongly suggest converting to Subversion's fsfs backend. Marius Gedminas -- As easy as 3.14159265358979323846264338327950288419716 signature.asc Description: Digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: [maemo-developers] python + osso_browser does work.
Hi, I'll put the readline.so in the python2.5-dev package. Installing this package will enable the readline functionalities in Python's interactive mode. Thanks, Osvaldo On 12/15/06, Jack Jansen <[EMAIL PROTECTED]> wrote: On 12-Dec-2006, at 17:15 , Osvaldo Santana wrote: Would it be possible to package readline.so separately so that Python developers can apt-get install python-readline, while regular users can save half a meg? It's possible to make another package with the readline.so module, but the Python interactive mode will not use it. Why is this? This isn't a problem on other platforms. For instance, on the Mac readline used to be a separate distribution because of licensing issues. -- Jack Jansen, <[EMAIL PROTECTED]>, http://www.cwi.nl/~jack If I can't dance I don't want to be part of your revolution -- Emma Goldman -- Osvaldo Santana Neto (aCiDBaSe) http://www.pythonologia.org ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
[maemo-developers] Re: mmc, one more time
> apt-get install e2fsprogs > This will install e.g. mkfs.ext2 and mkfs.ext3, as well as mkswap Thank you for advice! Cool! Mkswap is already installed on 770. Would it interact? I found e2fsprogs in repository, in pool directory. There are libraries, supporting this. What will be instal- led, in deed? Default has only tableteer. Additio- nal servers must be listed too. > This is all you need to get a formated partition with either ext2 (which I > use) or ext3. Huh! If ext3 support is in kernel, I'd take it. For ext2, module has to be loaded. Looking over directories, I've found sfdisk. Manual for it is a little bit cryptic. Expects something like: sfdisk -uM /dev/mmcblk0 < my.conf.file I'm not sure for "d" flag. This "my.conf.file" is something I should put to sfdisk. Syntax is real gem. Is there someone doing it before? My partitions on 1gb would be: 1. vfat (250mb, bootable) 2. swap (64mb) 3. ext3 (the rest of mmc) I'd like to have advices on this, prior to become owner of unusable mmc. Best regards Zoran ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
[maemo-developers] Creating Home Applet
I want to create a home applet (and task navigator entry) that can exist alongside the current web shortcut applet, but basically provide the same function. Even if someone could just point me in the right direction of the source code it would probably be enough. I basically want an applet that can start a new browser window and make it display fullscreen, but i can't work out how to start the browser with any kind of commandline options (if this is the way its done). I've found /usr/bin/browser starts the browser, but no options seem to be passed through this - it seem to be just a link to /usr/bin/maemo-invoker (?). Any help would be great, thanks Andrew ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: [maemo-developers] browser launch from shell script
tomi, have you tried url = "/etc/fstab " run-standalone.sh /usr/bin/browser --url=$url ? it works for me with sirocco On 12/16/06, Tomi Ollila <[EMAIL PROTECTED]> wrote: I've been searhing a way to launch browser from shell script, and this is the only way I've found (couple of references): --8<8<8<8<8<8<8<8<-- #!/bin/sh #open a url passed as $1 (found on maemo-developers ml) if ! dbus-send --system --type=method_call --dest="com.nokia.osso_browser" --print-reply /com/nokia/osso_browser/request com.nokia.osso_browser.load_url string:"$1" ; then DISPLAY=:0 run-standalone.sh /usr/bin/browser &>/dev/null & sleep 5 dbus-send --system --type=method_call --dest="com.nokia.osso_browser" --print-reply /com/nokia/osso_browser/request com.nokia.osso_browser.load_url string:"$1" fi --8<8<8<8<8<8<8<8<-- Is there a better way, or should I just use the following: --8<8<8<8<8<8<8<8<-- #!/bin/sh send_url () { dbus-send --system --type=method_call --dest=com.nokia.osso_browser \ --print-reply /com/nokia/osso_browser/request \ com.nokia.osso_browser.load_url string:"$1" && exit 0 } send_url "$1" DISPLAY=:0 run-standalone.sh /usr/bin/browser >/dev/null 2>&1 & for _ in 1 2 3 4 5 6 7 8 9; do sleep 1; send_url "$1"; done exit 1 --8<8<8<8<8<8<8<8<-- Tomi ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers -- --Antonio Gomes ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
RE: [maemo-developers] Taste the Herring!
> Is there some specific reason why instructions tell to select > CPU-transparency in the devkits when creating a i386 target? I think > that shouldn't be needed, but it should cause any problems > either since > no CPU-transparency method is selected. > No. It really should not be needed. ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers