(madtw @ april 21 2005,15:40 Wrote:first comment is that you probably should try a newer version of xbmc. there have been a bunch of changes (including bug fixes) to the xbmc python module in early april.

as for the error log you posted, the player errors probably happened after you tried to exit the live tv screen? when the live tv window is closed, the code first destroys the player object that was created for live tv playback... then the window should've closed. if it didn't then the player object would no longer exist and you would get a bunch of errors any time you tried to do anything that used the player object. i can change the way the code works so that it will recreate the player object if it doesn't exist or not delete the player object stored in the window (and let python clean it up when the window is destroyed) but if everything is working as it should, the player object should never be accessed after an attempted window close.

that last error is a bit puzzling... win had to exist for the domodal() call so why does it no longer exist. i remember seeing that error before (last year)... the fix at the time was to comment out the line.

(jpf @ april 21 2005,16:47 Wrote:

(madtw @ april 21 2005,15:40 Wrote:first comment is that you probably should try a newer version of xbmc. there have been a bunch of changes (including bug fixes) to the xbmc python module in early april.

as for the error log you posted, the player errors probably happened after you tried to exit the live tv screen? when the live tv window is closed, the code first destroys the player object that was created for live tv playback... then the window should've closed. if it didn't then the player object would no longer exist and you would get a bunch of errors any time you tried to do anything that used the player object. i can change the way the code works so that it will recreate the player object if it doesn't exist or not delete the player object stored in the window (and let python clean it up when the window is destroyed) but if everything is working as it should, the player object should never be accessed after an attempted window close.

that last error is a bit puzzling... win had to exist for the domodal() call so why does it no longer exist. i remember seeing that error before (last year)... the fix at the time was to comment out the line.

hi,

ok well basically the newer versions of xbmc have horrible bugs, the mplayer window shows a black screen sometimes amongst other random problems.

when i can't exit the screens, basically i can still use all the menus in the screen i'm stuck in. the main screen that freezes is the recordedshowdetails screen, but the livetv screen does that periodically as well. i also was having problems with connections freezing, and when i killed the mythbackend then i magically got control of the menus again, so i changed the socket to a timeoutsocket, put a default of 20 seconds in there, i haven't had a freeze like that since.

also, why was the "back" button disabled? i reenabled it here:
# stop user from quitting selecting parent dir in main window
if action == mythtvgui.action_parent_dir:
actionconsumed = none
maybe the back button will work even if the "menu" button doesn't, but maybe there is something i'm not aware of.

here is the change i made to def connect()
def connect( self ):
s = timeoutsocket.socket( socket.af_inet, socket.sock_stream )
timeoutsocket.setdefaultsockettimeout(20);

with regards to the last error, i've seen it a few times. i am still able to exit onto the main menu and then everything seems fine.

why is the program referencing a window id of -10000? that happens every time i can't exit the menu and am trying to by pressing the "menu" key.

it would probably be good to add code to recreate any connection that fails (connection reset by peer is an error i see sometimes), as well as recreate the player object if it does not exist.

(jpf @ april 21 2005,17:05 Wrote:

(jpf @ april 21 2005,16:47 Wrote:

(madtw @ april 21 2005,15:40 Wrote:first comment is that you probably should try a newer version of xbmc. there have been a bunch of changes (including bug fixes) to the xbmc python module in early april.

as for the error log you posted, the player errors probably happened after you tried to exit the live tv screen? when the live tv window is closed, the code first destroys the player object that was created for live tv playback... then the window should've closed. if it didn't then the player object would no longer exist and you would get a bunch of errors any time you tried to do anything that used the player object. i can change the way the code works so that it will recreate the player object if it doesn't exist or not delete the player object stored in the window (and let python clean it up when the window is destroyed) but if everything is working as it should, the player object should never be accessed after an attempted window close.

that last error is a bit puzzling... win had to exist for the domodal() call so why does it no longer exist. i remember seeing that error before (last year)... the fix at the time was to comment out the line.

hi,

ok well basically the newer versions of xbmc have horrible bugs, the mplayer window shows a black screen sometimes amongst other random problems.

when i can't exit the screens, basically i can still use all the menus in the screen i'm stuck in. the main screen that freezes is the recordedshowdetails screen, but the livetv screen does that periodically as well. i also was having problems with connections freezing, and when i killed the mythbackend then i magically got control of the menus again, so i changed the socket to a timeoutsocket, put a default of 20 seconds in there, i haven't had a freeze like that since.

also, why was the "back" button disabled? i reenabled it here:
# stop user from quitting selecting parent dir in main window
if action == mythtvgui.action_parent_dir:
actionconsumed = none
maybe the back button will work even if the "menu" button doesn't, but maybe there is something i'm not aware of.

here is the change i made to def connect()
def connect( self ):
s = timeoutsocket.socket( socket.af_inet, socket.sock_stream )
timeoutsocket.setdefaultsockettimeout(20);

with regards to the last error, i've seen it a few times. i am still able to exit onto the main menu and then everything seems fine.

why is the program referencing a window id of -10000? that happens every time i can't exit the menu and am trying to by pressing the "menu" key.

it would probably be good to add code to recreate any connection that fails (connection reset by peer is an error i see sometimes), as well as recreate the player object if it does not exist.

could you point me towards a xbmc-cvs build that actually works? and to a location. the primary -my mom- mirror only has a build since 2005-04-01

(madtw @ april 21 2005,04:01 Wrote:luigi: henk says your guide problem is fixed with my latest patches in the xbmcmythtv cvs repository.

(madtw @ april 22 2005,06:23 Wrote:henk: try the latest cvs...

(henk @ april 22 2005,07:33 Wrote:

(madtw @ april 22 2005,06:23 Wrote:henk: try the latest cvs...

the error was on the latest cvs

(madtw @ april 22 2005,06:23 Wrote:henk: try the latest cvs...

jpf: i intentionally made it so that the back & menu buttons work for subscreens but the menu button only works for the main screen. it is mainly so that you don't ditch out of the script unintentionally... i use the back button when closing subscreens and then hit the menu button when i truly want to quit the script from the main screen. to each his own i guess...

the timeout on the socket is a good idea.. i originally had the timeout coded when i was testing in linux but python in xbmc didn't like the code. in the end i ended up ripping it out. now that you've figured out the magic incantantation, i can put the socket timeout code back in...

when i couldn't exit the screens a while ago, i tried using other actions besides the back and menu buttons but it did not work. the more i read about your debugging, the more i think this might be a threading issue... one thread is blocked on a socket and the other thread requests the window be closed. since there are still pending events on the window, it isn't entirely freed up but new events cannot be created because the window id is no longer valid? the -10000 window id is something xbmc assigns to new python windows... i haven't hard coded any ids into the scripts.

if you take a look at mythtvguide.py, you can see how i restricted threaded events from modifying the screen at the same time. you could try adding in similar code into the screens that are causing problems to see if it helps... start at line 489.

as for recreating connections... with some quick calls to test if a connection is still valid, it should be possible to add in recreating a dead connection into the getinstance calls that retrieve a connection/database object. at the moment, i try to hang on to connections for as short a time as possible but there is no retry. this has been working well for me so far but i must not be getting any timeouts on my connections.

i'm not sure what the problem is with your last mysql error... i haven't touched that module in a long time. what version of mysql are you using? i'm running version 4.0.23... if you're running a newer version it could be that something has changed breaking mysql.py.

(jpf @ april 22 2005,13:10 Wrote:ok well, the thing is that the screens don't freeze. i just can't exit them. for example in the recorded movies screen, i often get stuck in the preview window. i can play the movie from there, i can delete and copy etc. i just can't exit it. can we put an emergency exit key in there? is there some special way that the program needs to be killed completely?

(madtw @ april 23 2005,04:58 Wrote:

(jpf @ april 22 2005,13:10 Wrote:ok well, the thing is that the screens don't freeze. i just can't exit them. for example in the recorded movies screen, i often get stuck in the preview window. i can play the movie from there, i can delete and copy etc. i just can't exit it. can we put an emergency exit key in there? is there some special way that the program needs to be killed completely?

yeah, i know exactly what you're talking about. i haven't experienced it in a few months now. when i had the problem, i tried putting in emergency exit buttons, etc. to get around it. nothing worked... it seemed like python lost all concept of the window (as far as closing it goes) but the events in the window would still fire (e.g. pushing buttons, etc.).

(jpf @ april 23 2005,12:26 Wrote:

(madtw @ april 23 2005,04:58 Wrote:

(jpf @ april 22 2005,13:10 Wrote:ok well, the thing is that the screens don't freeze. i just can't exit them. for example in the recorded movies screen, i often get stuck in the preview window. i can play the movie from there, i can delete and copy etc. i just can't exit it. can we put an emergency exit key in there? is there some special way that the program needs to be killed completely?

yeah, i know exactly what you're talking about. i haven't experienced it in a few months now. when i had the problem, i tried putting in emergency exit buttons, etc. to get around it. nothing worked... it seemed like python lost all concept of the window (as far as closing it goes) but the events in the window would still fire (e.g. pushing buttons, etc.).

what we need to figure out is how my setup differs from yours. do you try various cvs versions of xbmc? i have tried many versions and always have the same problem.

you are obviously always using the latest cvs so there shouldn't be a code difference there.

i use international characters though, i wonder if that could have an effect?

(lintweaker @ april 24 2005,20:17 Wrote:mythtv 0.18 is working beautifully with the latest xbmcmythtv from cvs. i am using it with an "older" version of xmbc. newer versions have some anoying bugs.

the only thing is that in the program guide shows with long names are overlapping. no biggy ...

excellent stuff.
:kickass: