inside releasechecker one can find time.sleep(0.5) surely there must be a better way than that and short of that I'll suggest that the sleep be increased to something less resource consuming (5 seconds seems like a good enough thing). Reproducible: Always
Git commit d6181540fa9a76b01526af0f0eced81dc80d0d2c by Harald Sitter. Committed on 04/11/2013 at 14:58. Pushed by sitter into branch '2.1'. Don't busy wait in releasechecker .download() is a sync interface, so use that instead. Also remove pointless debug. FIXED-IN: 2.1.1 M +1 -4 kded/distupgradeevent/releasechecker http://commits.kde.org/muon/d6181540fa9a76b01526af0f0eced81dc80d0d2c
Git commit ed6187b74b38d5b26c07e1a0a33a468dc00b573d by Harald Sitter. Committed on 05/11/2013 at 16:07. Pushed by sitter into branch '2.1'. Reintroduce a busy-wait to releasechecker and document why! Unfortunately due to incredibly crappy design of MetaReleaseCore we cannot avoid the pointless context switches to check for .downloading. As download() is started in a thread and we have no external control over the thread, it is autostarted in __init__ and there are no synchronization methods provided by MetaReleaseCore. Calling .download() manually which would provide a nicely sync function will result in unprotected threading leading to incorrect upgrade notifications again. M +13 -1 kded/distupgradeevent/releasechecker http://commits.kde.org/muon/ed6187b74b38d5b26c07e1a0a33a468dc00b573d