On Thu, Dec 07, 2000 at 11:12:48AM -0500, Christopher C. Chimelis wrote: > > On Thu, 7 Dec 2000, Stuart Anderson wrote: > > > I seemed to miss the patch mentioned, but my guess would be that it involves > > wether or not to pass -s to install? > > > > The modules that get loaded can't get completely stripped, or they loose the > > symbol & relocation information which is required for loading. It is probably > > OK to strip out some other stuff like debugging info. > > > > I just peeked at the config file, and I don't see where -s is getting set. Is > > this a patch to another patch? Has the default behavior of install changed? > > Should I just wait for the coffee to kick in and then all this will make sense? > > Not really...it's not -s that's the problem, it's dh_strip. In the patch, > I placed an ifeq in the rules file that made it so that, on Alpha, > dh_strip is called with '-Nxserver-xfree86', but on other archs, it > isn't. Stuart may not be familiar with the vagaries of Debian packaging. I build with debugging symbols enabled, but Debian Policy insists that they be stripped out when the package is generated. My completely uneducated guess is that something in module loader "knows" that debugging symbols are enabled and depends on this, so that when later this same code is run stripped, it becomes confused? -- G. Branden Robinson | We either learn from history or, Debian GNU/Linux | uh, well, something bad will happen. branden@deadbeast.net | -- Bob Church http://deadbeast.net/~branden/ |
Attachment:
pgp7PSEWj8Rk7.pgp
Description: PGP signature