Frequent Hiccups / Stalls on Mac OS X
Status: Alpha
Brought to you by:
petrocket
Menu
▾
▴
#31 Frequent Hiccups / Stalls on Mac OS X
One issue I'm running into with the current Mac OS version is that it has pretty severe hiccups. I tried it on my Late 2012 iMac as well as a rather recent MacBook Pro, both running on 10.9.5 (actually the MacBook runs on 10.9.4). Both with fast processors and plenty of memory (32GB, 16GB; 3.4 GHz Intel Core i7, 2.6 GHz Intel Core i7). I also tried this after rebooting to make there was no interferences with any other apps and kept it in the foreground to make sure it was no issue with App Nap (which for some reason I cannot disable for Spacescape).
Basically, the behavior I'm seeing is that all works fine and smooth but almost all the time I click into any of the properties, it gets stuck for a few seconds (even before changing anything). Then, when I do change something, usually it's quite responsive but not always. In other words: Most of the time I work on Spacescape I spend waiting.
Events that cause hiccups (sporadically, i.e. not always but very frequently): Selecting a layer; opening a layer; selecting a property; changing a property.
It gets worse with more complex spacescapes ... then sometimes I have to wait up to a minute until Spacescape becomes responsive again.
Note: The Windows version does not have those hiccups (even when running in VMWare Fusion ;-) ). In other words: I cannot reproduce this when running the Windows version in Windows 8.1 running via VMWare Fusion. But there, I have another problem (see https://sourceforge.net/p/spacescape/bugs/32/).
Bummer! I'll see if there is something I can do to help avoid the hangs, unfortunately, I suspect this has to do with the Qt framework and less with Spacescape since the code is pretty much identical otherwise.
I'm not sure if this is still applicable but I believe a long time ago, Mac OS had a different approach for multithreading (cooperative where the thread has to say "ok, now other threads can continue working" instead of a scheduler taking control away from threads when it considered it a good point in time to do so ;-) ).
The behavior I'm seeing does look a little bit like something potentially going wrong there, like maybe the GUI thread giving control away and not getting it returned (it appears to really just be a GUI issue). But this could very well be an issue with QT and anyways it's just a kind of wild guess.