GUI is not working...

Using (testing) v1.1.5 Using on Ubuntu 20.04

From the GUI

Still seeing this issue

Debugging this with the command line given here it is working.

Maybe the issue is with the flag Open spreadsheet So, if I unmark this checkbox, the whole gui hangs.

A 3rd test gave this… BOM file: board.csv Reconigzed EDAs: CSV file Open spreadsheet unmarked

[3]    76634 segmentation fault (core dumped)  kicost

[sliedes]

Ok, I believe a few issues are conflated here (especially since #421 was marked as a duplicate). I think I have an idea what’s going on.

1. The issue in https://github.com/xesscorp/KiCost/issues/421#issuecomment-691255711: wxPython has become more strict about accepting nonsensical combinations of sizer flags. From the release notes of wxPython 4.1.1:

This should have been mentioned in the notes for the last release, but alas, it wandered away and got lost. wxWidgets is now validating the flags passed when adding items to a sizer, to ensure that they are the correct flags for the type of the sizer. If the given flags do not make sense, for example using horizontal alignment flags in a horizontal box sizer, then a wxAssertionError error is raised.

… which is exactly what happens. The error messages even tell what the problem is (“wxEXPAND flag will be overridden by alignment flags”). The solution is to either remove the wx.EXPAND flag or the alignment flags, depending on what the intent here is. Getting the GUI to start required a modification in kicost_gui.py of lines 367, 384 and 449. I’m not sending a pull request since I don’t know what the GUI is supposed to look like and hence which of the conflicting flags should be removed.

1. The xcb/XInitThreads error. What is happening here is that GUI code is being run from a thread different than the one that created the window. This is not supported in wxPython. Specifically, it seems that formKiCost.run() is run from a different thread (started in button_run()), and it contains GUI code such as self.m_gauge_process.SetValue(0) (and lots of other GUI manipulation). This causes breakage. See eg. this thread. Hacking button_run() to just call self.run() instead of starting a thread fixes this for me.

[set-soft]

Hi @hildogjr !

If you incorporate the flake8 patch first (which I strongly recommend) the #451 becomes:

diff --git a/kicost/kicost_gui.py b/kicost/kicost_gui.py
index 231e11f..bb2dcb9 100644
--- a/kicost/kicost_gui.py
+++ b/kicost/kicost_gui.py
@@ -41,7 +41,7 @@ import webbrowser  # To update informations.
 import sys
 import os
 import subprocess  # To access OS commands and run in the shell.
-import threading
+from threading import Thread
 import time  # To elapse time.
 import tempfile  # To create the temporary log file.
 from datetime import datetime  # To create the log name, when asked to save.
@@ -96,6 +96,18 @@ PAGE_POWERED_BY = 'https://kitspace.org/'
 kicostPath = os.path.dirname(os.path.abspath(__file__))  # Application dir.
 
 
+# ======================================================================
+class ChildThread(Thread):
+    """ Helper class to safetly call the run action """
+    def __init__(self, myframe):
+        """Init Worker Thread Class."""
+        Thread.__init__(self)
+        self.myframe = myframe
+
+    def run(self):
+        wx.CallAfter(self.myframe.run)
+
+
 # ======================================================================
 def open_file(filepath):
     '''@brief Open a file with the default application in different OSs.
@@ -714,10 +726,9 @@ class formKiCost(wx.Frame):
     def button_run(self, event):
         ''' @brief Call to run KiCost.'''
         event.Skip()
-        # self.run()
-        # wx.CallLater(10, self.run) # Necessary to not '(core dumped)' with wxPython.
-        t = threading.Thread(target=self.run)  # , args=[self])
-        t.start()
+        self.child = ChildThread(myframe=self)
+        self.child.daemon = True
+        self.child.start()
 
     # ----------------------------------------------------------------------
     # @anythread