do not mistake WAD-named directories for actual WAD files

[fabiangreffrath]

Fixes a segfault when running the engine e.g. from the savegames directory.

Does anyone know the reason why we do not simply return false here?

woof/src/m_misc.c

Lines 50 to 56 in ec7344f

	else
	{
	// If we can't open because the file is a directory, the
	// "file" exists at least!
	return errno == EISDIR;
	}

[fabiangreffrath]

Does anyone know the reason why we do not simply return false here?

I mean, do we ever use M_FileExists() to check for the existence of a directory?

[fabiangreffrath]

I mean, do we ever use M_FileExists() to check for the existence of a directory?

Even worse, POSIX fopen() returns a valid FILE * pointer for a directory. How is it on Windows?

[fabiangreffrath]

Alternative approach:

--- a/src/m_misc.c
+++ b/src/m_misc.c
@@ -45,14 +45,11 @@ boolean M_FileExists(const char *filename)
     if (fstream != NULL)
     {
         fclose(fstream);
-        return true;
+        return M_DirExists(filename) == false;
     }
     else
     {
-        // If we can't open because the file is a directory, the
-        // "file" exists at least!
-
-        return errno == EISDIR;
+        return false;
     }
 }
 

[rfomin]

I mean, do we ever use M_FileExists() to check for the existence of a directory?

Even worse, POSIX fopen() returns a valid FILE * pointer for a directory. How is it on Windows?

I can't test it at the moment (winter holidays).

I like the PR, the only issue is that it would be different from Choco, so maybe we should rename the M_FileExists() function somehow.

[fabiangreffrath]

I like the PR

Do you mean the alternative approach that I posted before? I'll integrate it better into the rest of the code.

[rfomin]

I like the PR

Do you mean the alternative approach that I posted before? I'll integrate it better into the rest of the code.

Yes, alternative appoach is fine too.