Infinite loop of errors under GTK<4.10 after recent commit about gtg HOT 6 OPEN

It seems that I had the entire thing before - this is the start of the log:

2024-04-29 08:48:03,485 - WARNING - __init__:__init__:68 - Backend backend_caldav could not be loaded: ModuleNotFoundError("No module named 'caldav'")
Gtk-Message: 08:48:03.674: GtkDialog mapped without a transient parent. This is discouraged.
Traceback (most recent call last):
  File "<< GTG source >>/.local_build/install/lib/python3/dist-packages/GTG/gtk/browser/sidebar.py", line 286, in tags_setup_cb
    expander.set_indent_for_depth(True)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'TreeExpander' object has no attribute 'set_indent_for_depth'. Did you mean: 'set_indent_for_icon'?
Gtk-Message: 08:48:03.678: GtkDialog mapped without a transient parent. This is discouraged.
Traceback (most recent call last):
  File "<< GTG source >>/.local_build/install/lib/python3/dist-packages/GTG/gtk/browser/sidebar.py", line 355, in tags_bind_cb
    expander = box.get_first_child()
               ^^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'get_first_child'

and then everything repeats except the first line.

So, two things I'd like to do here:

• Require GTK 4.10 explicitly
• Avoid opening an exception dialog when one is already open

from gtg.

I am trying to reproduce the crash. Could you please provide more details about your setup? Does this happen with the default dataset? What is your exact GTK version?

There are two different errors:

• AttributeError: 'TreeExpander' object has no attribute 'set_indent_for_depth'
  Based on the GTK documentation, this method is present from version 4.10. Is it possible that you have an older version?

• AttributeError: 'NoneType' object has no attribute 'get_first_child'
  Maybe this is just a consequence of the first error. Could you please try to capture the beginning of your error message?

from gtg.

Thanks for looking at it! I have GTK 4.8.3, which seems to explain the problem. Let's make 4.10 an explicit requirement then (here, I think? and in the README).

It happens with the default dataset as well.

Could you please try to capture the beginning of your error message?

I'll try that when I have a minute! I assume I just need to disable the code that catches the exception and opens the dialog.

Side note: I'll probably create new issue for this, but I'd like to avoid "infinite dialogs" as a failure mode. Probably by detecting whether the error dialog is already open, and avoiding to open a new one.

from gtg.

I'll try that when I have a minute! I assume I just need to disable the code that catches the exception and opens the dialog.

You can also just redirect the error output to a file like this:

./launch.sh 2> output.txt

Then hit Ctrl + C in the terminal to exit GTG.

from gtg.

Some pieces of notes with my current understanding of the problem:

1. The TaskPane is created
2. It connects some "setup" signal to task_setup_cb()
3. task_setup_cb() is called
4. It tries to create a TaskBox
5. The constructor of TaskBox calls self.expander.set_indent_for_depth(True), which raises an AttributeError in GTK<4.10
6. task_setup_cb() propagates the exception
7. It gets caught by the application, which creates an opens the ExceptionHandlerDialog
8. Meanwhile, the "setup" signal is sent again, and we're back at step 3

Being on GTK 4.10 means that step 5 doesn't fail, but any exception in the TaskBox constructor has the same effect.

My current idea is to not open the ExceptionHandlerDialog of step 7 if one is already open.

from gtg.