[mythtv] [mythtv-commits] Ticket #4381: Using a Miscellaneous Status Information (see ticket #4049) makes mythbackend segfault occasionnally...

[Michael T. Dean]

On 04/12/2010 11:45 PM, Nicolas Riendeau wrote:
> Michael T. Dean wrote:
>
>> So it took me a while (just under 2 1/4 years :) to get around to it, 
>> but today--on the day I deleted the myth_sensors.sh script from SVN ( 
>> http://svn.mythtv.org/trac/changeset/24055 )--I went ahead and 
>> updated the script to be a Perl-based script (based on the one I 
>> started in 2008) which may work properly, regardless of locale.  The 
>> updated version is at http://www.mythtv.org/wiki/Myth_sensors.pl .  
>> Please feel free to try it out, and--if it doesn't work--feel free to 
>> update/enhance it.  :)
>
> Hi Mike!
>
> I'll try it as soon as I get a chance and let you know the results.
>
> BTW, didn't you say back then that you wanted to put something in 
> place so that the script would be killed if it ran for too long (or is 
> that just my imagination? (-; ).

You weren't supposed to remember that part of the conversation.

> Would there be any way to do it in Perl now that the script is in Perl?

The original implementation of the Miscellaneous Status Information used 
QProcess with a 10s timeout to run the script--so if the user ran a 
script that took too long, it would stop waiting after 10s and ignore 
the script.  Unfortunately, I later found out that Qt3 QProcess could 
only be run from the event thread (and caused occasional backend crashes 
when the backend used it for the Misc Status Info), so I converted it to 
use popen(), but in doing so had to lose the 10s timeout.  I had 
mentioned that with Qt4, we'd likely be able to return to using QProcess 
and reinstate the 10s timeout.  That said, it's only been about 2 years 
and 3 weeks since we switched to Qt4, so I haven't gotten a chance to 
revisit that, yet.  :)

>
> Thank you and have a nice day!

Thank you for the testing (then and future).  :)

Mike