I don't have commit access so someone will have to merge this for me. Thanks :-)

I don't have commit access but please let me know if anything else is needed before it can be merged.

I have updated the description to tag relevant bugs

as I don't have a developer account, I will need someone else to push it please :-)

in addition, just returning early instead of the assert causes a different bug, which is that compilerEdited gets called when we select out of the compiler, due to the same kLineEdit bug that causes textEdited to be emitted on calling clear().

In D9809#189484, @mwolff wrote:

So just remove the assert - instead return early when indexes.isEmpty(). That way we don't need to mess with the signal blocker and the code behaves as intended. And you remove code instead of adding more.

improved backtrace with all symbols now attached to the bug report:
https://bugs.kde.org/show_bug.cgi?id=373004

In D9808#189467, @dfaure wrote:

Sorry to be a pain, but can you also update the description, which still talks about textEmitted()?

(FYI using arc diff to upload the patch would also provide context, but OK, I can open the file locally to find out about the context for those unittest changes....)

as per review, this updated diff uses QSignalBlocker instead of blockSignals()

from the git history I believe this issue has been in place since before 5.0.

this updated diff removes the rename, just has fixed and extended unit tests and the one line fix for not emitting the textChanged signal spuriously.

In D9809#189240, @mwolff wrote:

I don't understand how this fixes the crash. Can you extend the bug report with a better backtrace that contains more symbols? Also, the bug report looks like it's due to a queued signal emission. Which signal is that? And it looks like something withing the kdevplatform project library, how is that related to the compiler selection?