Distro: Kubuntu 16.04 KDE Plasma: 5.8.5 KDE Frameworks: 5.28.0 QT: 5.6.1 Linux: 4.9.13 Arch: amd64 KDevelop: KDevelop-5.1.0-x86_64.AppImage Since KDevelop-5.0.3-x86_64.AppImage (5.0.2 is OK) I observe following buggy behavior of KDevelop when editing Python sources. After several minutes of active coding, KDevelop starts eating 100 % of one CPU core. Besides the CPU load IDE seems to work just fine. When "Rename variable" dialog is opened the CPU load stops and starts again when dialog is closed. It is usually not possible to stop KDevelop by simple closing the GUI window. CPU consuming process keeps running and needs to be killed manually. I am not sure but it seems that this happens only when Vi Input Mode is selected and when editing Python project. I didn't observe this problem when tried to use KDevelop for C/C++ sources (with Vi input enabled) and for Python sources (with Vi input disabled) but it might be just fortune.
Does that always happen for you? Could you share the Python project you're working on?
(In reply to Kevin Funk from comment #1) > Does that always happen for you? Could you share the Python project you're > working on? This happens always but after various amount of time (up to tens of minutes). I am sorry but I can't share the code since this is a commercial project. However I was able to reproduce the bug on shadowsocks code https://pypi.python.org/pypi/shadowsocks-c/2.8.2
The VI mode sounds like a random correlation to me. I don't see how that could have anything to do with it. Can you attach the debugger to KDevelop when it uses too much CPU, and then interrupt it (Ctrl+C) and post the output of "thread apply all bt"? Also, are you sure it's not just the background parser doing something and will eventually finish after a few seconds ...?
(In reply to Sven Brauch from comment #3) The VI mode might be just coincidence but I didn't observe the CPU load while using KDevelop without VI mode for several hours. I can attach the debugger but the debug symbols seems to be stripped from kdevelop binary in provided appimage. Moreover the debugger (running under root account) returns error message like "Backtrace stopped: Cannot access memory at address 0xc8a341b0" for all the threads. Any other suggestions regarding the debugger? I am sure that if the CPU load problem starts it does not stop until the KDevelop is closed (I have tested this for several hours). I am pretty sure that the problem appeared with KDevelop-5.0.3-x86_64.AppImage and that the 5.0.2 is OK.
Hm, I loaded that project here and didn't see any problems when clicking around for a while. One thing you can try is closing KDevelop, then removing ~/.cache/kdevduchain -- does that help? If not, I don't think you can get symbols from the AppImage, eventually we should provide a debug build in the future ... but maybe even a trace without full symbols would be useful.
I have already tried to start new project (which should be the same as removing ~/.cache/kdevduchain, I hope) with no luck. Please try to simulate the problem with VI mode enabled. I did not observe the problem with VI mode disabled. As I mentioned in my previous comment I am unable to run the backtrace receiving messages like: "Backtrace stopped: Cannot access memory at address 0xc8a341b0"
It seems that this is not python related issue. The IDE behaves the same way even if editing C source code with enabled VIM plugin.
The bug seems to be related to AppImage only. I'm using KDevelop 5.1.1 from Kubuntu-ppa Backports repository without that problem.
I don't think there's enough information to determine the issue here, or recognise if it was fixed. The AppImage builds have been changed quite significantly since 5.1.0. If you have a similar problem with a current build, please reopen or file a new bug.