Re: loader: fix loadorder

[Alexandre Julliard]

"Dimitrie O. Paun"  writes:
> ChangeLog
>   Revert default loadorder to native DLLs firsts, to use
>   the programs supplied ones if available.
>   (Bug discovered and investigated by Tony Lambregts)

I don't think we should revert it, using builtins by default is a good
idea IMO. We just need to handle dlls with an explicit path
differently; how about something like this?

Index: loader/loadorder.c
===================================================================
RCS file: /opt/cvs-commit/wine/loader/loadorder.c,v
retrieving revision 1.60
diff -u -r1.60 loadorder.c
--- loader/loadorder.c	23 Oct 2002 20:20:59 -0000	1.60
+++ loader/loadorder.c	14 Feb 2003 19:24:14 -0000
@@ -103,6 +103,12 @@
     LOADORDER_BI, LOADORDER_DLL, 0, 0
 };

+/* default for modules with an explicit path */
+static const enum loadorder_type default_path_loadorder[LOADORDER_NTYPES] =
+{
+    LOADORDER_DLL, LOADORDER_BI, 0, 0
+};
+
 static struct loadorder_list cmdline_list;

@@ -674,13 +680,18 @@
                    debugstr_loadorder(loadorder), debugstr_a(path) );
             goto done;
         }
-    }
-
-    /* and last the hard-coded default */
-    memcpy( loadorder, default_loadorder, sizeof(default_loadorder) );
-    TRACE( "got hardcoded default %s for %s\n",
-           debugstr_loadorder(loadorder), debugstr_a(path) );

+        /* and last the hard-coded default */
+        memcpy( loadorder, default_loadorder, sizeof(default_loadorder) );
+        TRACE( "got hardcoded default %s for %s\n",
+               debugstr_loadorder(loadorder), debugstr_a(path) );
+    }
+    else  /* module contains an explicit path */
+    {
+        memcpy( loadorder, default_path_loadorder, sizeof(default_path_loadorder) );
+        TRACE( "got hardcoded path default %s for %s\n",
+               debugstr_loadorder(loadorder), debugstr_a(path) );
+    }

done:
     if (app_key) NtClose( app_key );

loader: fix loadorder

[Dimitrie O. Paun]

On February 14, 2003 02:26 pm, Alexandre Julliard wrote:
> I don't think we should revert it, using builtins by default is a good
> idea IMO. We just need to handle dlls with an explicit path
> differently; how about something like this?

I agree. My patch worked by the virtue of the [DllOverrides], and the
hard coded defaults. But what if things change in ~/.wine/config, or
there's no such file? To be honest, I knew this is a portion of code
that you worked on a lot, so I just sent in the patch to get your
attention... :)

Tony should give it a try, and see if it works as expected.

Thanks for the patch!

loader: fix loadorder

[Tony Lambregts]

Dimitrie O. Paun wrote:
> Tony should give it a try, and see if it works as expected.

I just did. Wonderfull.
> Thanks for the patch!

Ditto. Alexandre, could please apply it. (Pretty please)

loader: fix loadorder

[Rein Klazes]

On 14 Feb 2003 11:26:32 -0800, you wrote:
> > Revert default loadorder to native DLLs firsts, to use
> >   the programs supplied ones if available.
> >   (Bug discovered and investigated by Tony Lambregts)
>
> I don't think we should revert it, using builtins by default is a good
> idea IMO. We just need to handle dlls with an explicit path
> differently; how about something like this?

It fixes girotel, which has a comm.dll in its own directory that clashes
with builtin comm.drv.

Rein.