fix bugs with pyexists and qt path #34
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #34 +/- ##
=======================================
Coverage ? 19.51%
=======================================
Files ? 5
Lines ? 451
Branches ? 0
=======================================
Hits ? 88
Misses ? 363
Partials ? 0 ☔ View full report in Codecov by Sentry. |
| fixqtpath() | ||
| # `fixqtpath()` seems to not be working, | ||
| # https://github.com/JuliaPy/PythonPlot.jl/issues/17 | ||
| #fixqtpath() |
There was a problem hiding this comment.
It is it not working, or not necessary? What is the downside of keeping this function call (which may still be needed on some systems)?
There was a problem hiding this comment.
It somehow breaks the Qt5 backend for me, and apparently others as described in #17 . It isn't necessary for me; I don't even need to set QT_PLUGIN_PATH. After spending a few hours debugging the first problem, I didn't take time to investigate what is going here with qt path.
|
Any chance to get this merged, or what is missing? |
|
Bump. I was hoping to share code with colleagues that uses PythonPlot, and it would be easier if this is merged. |
On my system (MacOS with macports-installed julia and python), PythonPlot was not picking up an appropriate gui backend. It took some sleuthing to determine why, but I discovered that
pyimport("importlib")would sometimes not includeutilas an attribute. This PR fixes that problem, and also adds output of the error in the gui-detection try block to help with any future bugs that might arise there.I also include in this PR a change to comment out
fixqtpath()as it does not seem to work properly, as also reported in #17. I have not spent time chasing this one down. This change enables Qt5 backend on my system, even without specifyingENV["QT_PLUGIN_PATH"].