Closed danwalmsley closed 9 years ago
Unable to reproduce on current build, but will keep open and look out for this in the coming days.
Found the bug, need to reparse the file, marking TU as dirty and using this to indicate.
This has pointed out that the run diagnostics method uses up far too much time on the UI thread, this will need to be fixed, raising another issue.
This issue will be left open until that is fixed, and to keep an eye on the fix implemented here.
Still cropping up, will wait for #42 to be completed as this will re-model how this works.
I think this is caused by diagnostics still running when it triggers a re-run, then it exits straight away. We need to rethink the cancellation strategy.
Some kind of feedback when cancellation has been accepted, etc Or a task that can be instantly terminated (could be unsafe?)
Perhaps queue up runs and Clear the queue once new ones are added. I.e.only ever allowed a single queue waiting. Could actuallly be a flag. Clearwd at begining of run, if it is true at end of cancel Or completion tjen runagain.
This has now finally been fixed.
As you type errors are indicated, but sometimes it doesn't re-check the last change.