Assertion failed in wx_chdir on Windows 11

[rmunn]

Just installed UCBLogo version 6.2.5 on a Windows 11 computer. When launching it, I got the following error message:

Assertion failed!

Program: C:\Program Files (x86)\UCBLogo\ucblogo.exe
File: wxTerminal.cpp
Line: 2738

Expression: success

Line 2738 in the error message is line 2741 in the source:

ucblogo-code/wxTerminal.cpp

Lines 2739 to 2742 in 2d4140c

	extern "C" void wx_chdir(char *file_path) {
	bool success = wxSetWorkingDirectory(wxString::FromAscii(file_path));
	assert(success);
	}

I haven't yet been able to figure out what file_path wx_chdir is being called with here. But my Windows 11 installation is pretty standard: home directory under C:\Users, no spaces in my username so my home dir is C:\Users\rmunn, and so on.

The same assertion fails if I install UCBLogo for current user only rather than as administrator.

[jrincayc]

Hm, we probably should do something like this:

diff --git a/wxTerminal.cpp b/wxTerminal.cpp
index 0e3c89e..c12c67e 100644
--- a/wxTerminal.cpp
+++ b/wxTerminal.cpp
@@ -2738,5 +2738,8 @@ extern "C" char * wx_get_current_dir_name(void) {
 
 extern "C" void wx_chdir(char *file_path) {
   bool success = wxSetWorkingDirectory(wxString::FromAscii(file_path));
+  if(!success) {
+    fprintf(stderr,"Failed chdir '%s'\n", file_path);
+  }
   assert(success);
 }

so we have a better idea what is going on. One possibility is that file_path contains some characters that are not 7-bit ascii.

[rmunn]

If file_path is ASCII, then wxString::FromAscii(file_path) could be replaced with wxString::FromUTF8(file_path) with identical results. If file_path contains non-ASCII characters, then one of two things is likely:

• They are in UTF-8, and FromUTF8 will succeed.
• They are NOT in UTF-8, and FromUTF8 will fail with 99.99% probability.

If a file path is in CP-1252 / Latin-1 or some other charset, it is vanishingly unlikely that it will happen to be a valid UTF-8 string: most people just don't put the character sequence é in their filenames, for example. So the sequence 0xC3 0xA9 0x63 0x6rf 0x6c 0x65 could either decode as "école" (French for "school"), in UTF-8, or it could decode as "école" in CP-1252. But the sequence 0xE9 0x63 0x6rf 0x6c 0x65 ("école" in CP-1252) would be invalid UTF-8 and would produce an empty wxString according to the wxWidgets documentation.

So it should be safe to write a function to decode file paths, and other strings of unknown encoding, by the following heuristic:

• If input is empty, return empty wxString
• Try using wxString::FromUTF8
• If FromUTF8 returns non-empty wxString, return it as the result
• If FromUTF8 returns empty wxString, try using wxString::wxString(const char *) which assumes the char * is in the current Windows locale (which was most likely set by the Windows installer when it asked the user what language they want to use Windows in).

Then that function (let's call it toWxString for this example) would be used as follows:

diff --git a/wxTerminal.cpp b/wxTerminal.cpp
index 0e3c89e..c12c67e 100644
--- a/wxTerminal.cpp
+++ b/wxTerminal.cpp
@@ -2738,5 +2738,8 @@ extern "C" char * wx_get_current_dir_name(void) {
 
 extern "C" void wx_chdir(char *file_path) {
-   bool success = wxSetWorkingDirectory(wxString::FromAscii(file_path));
+   bool success = wxSetWorkingDirectory(toWxString(file_path));
+  if(!success) {
+    fprintf(stderr,"Failed chdir '%s'\n", file_path);
+  }
   assert(success);
 }

[rmunn]

https://docs.wxwidgets.org/stable/classwx_string.html#a2ddc1b7c8e1eb9adbf5874dead5b180b is my source for the fact that FromUTF8 returns an empty wxString if the input was not valid UTF-8.