<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.26.0">
</HEAD>
<BODY>
On Wed, 2010-09-22 at 09:16 -0700, Jeremy Huddleston wrote:<BR>
<BLOCKQUOTE TYPE=CITE>
<TT><FONT COLOR="#1a1a1a">Just to be 100% complete here... This has the effect that DRI_DRIVER_PATH="" rather than ${libdir}/dri on systems without dri.pc</FONT></TT><BR>
<BR>
<TT><FONT COLOR="#1a1a1a">Is that situation ok? Should you check for existence of dri.pc first and set it to the old fallback if not present?</FONT></TT><BR>
<BR>
</BLOCKQUOTE>
<BR>
I just check where used.<BR>
<BR>
<BLOCKQUOTE>
<PRE>
./glx/glxdriswrast.c:static const char dri_driver_path[] = DRI_DRIVER_PATH;
./glx/glxdri2.c:static const char dri_driver_path[] = DRI_DRIVER_PATH;
./glx/glxdri.c:static const char dri_driver_path[] = DRI_DRIVER_PATH;
</PRE>
</BLOCKQUOTE>
<BR>
Assuming this code is invoked if DRI is missing, it uses the full path to load the driver.
<BLOCKQUOTE>
<PRE>
snprintf(filename, sizeof filename,
"%s/%s_dri.so", dri_driver_path, driverName);
screen->driver = dlopen(filename, RTLD_LAZY | RTLD_LOCAL);
</PRE>
</BLOCKQUOTE>
I cannot be 100% sure that it won't uncover a dormant bug. If it does, it will be in xorg<BR>
code (as opposed to a dependency) and is fixable.
</BODY>
</HTML>