- From: David Schleef <ds_at_schleef.org>
- Date: Mon, 4 Aug 2003 15:55:43 -0700
On Mon, Aug 04, 2003 at 11:11:14PM +0100, Bernd Porr wrote: > Yes, I read it. However, it contradicts a bit with the last but one > paragraph where you say that it should be no problem to handle kernel > 2.6 modules. As a non native a need a bit more clearer expressions. > Maybe you could simply say what's going on. I said: > Agreed. I did a lot of work on this a few weeks ago as part of > a transition to using automake, but never checked it in because > it never fully worked (and wasn't even remotely well-tested.) "I did a bunch of work to replace the system." > There's nothing fundamentally difficult with supporting 2.6-style > kernel modules in the existing makefiles, however, I certainly > don't want to touch the steaming pile (which was derived from > Linux-2.2.17). "It could be possible to continue to use the existing system, but I hate it." > That system has lasted so long because it is > nearly bug free on 11 architectures, 3 kernel series, a variety > of random kernel configuration options, cross-compilation, and > voodoo linker flags. Any potential replacement system will need > to meet those requirements before being considered. "A replacement system is a hard problem." The last paragraph was specifically written to encourage people not to blindly start proposing fixes without understanding the problem and/or complain that a solution doesn't exist yet. dave...
Received on 2003-08-04Z21:55:43