Re: too much untested code in new kernels

Jes Degn Soerensen (jds@kom.auc.dk)
03 Jan 1997 10:39:53 +0100


>>>>> "Ray" == Ray Van Tassle-CRV004 <Ray_Van_Tassle-CRV004@email.mot.com> writes:

Ray> Hence my request for Linus (or _somebody) to make a test compile
Ray> with every possible option turned on, to make sure the d*mn thing
Ray> will COMPILE at least!

Ray> Whenever anybody mentions this, they get shouted down, claiming
Ray> (validly) that it is impossible for anybody to test all possible
Ray> combinations of options. But for CRYING OUT LOUD how about just
Ray> making sure it compiles!!!!!!!! And verifying that the resultant
Ray> image boots up.

And while we are at it, someone should also sponsor Linus a Sparc, a
m68k, a PPC and an ARM so Linus can compile and test on those
architectures as well. Forget it, your request is simply unrealistic!

You must remember drivers may not fail on some systems but will fail
on others because the hardware is detected incorrectly or drivers are
initialized in the wrong order. This is especially valid for
architectures such as the x86 where the hardware detection is pretty
idiotic (well except for PCI devices).

As long as we are dealing with development kernels it is more
important to get new things and bugfixes into the kernel. With the
rapid development we risk that a patch might break something else, but
thats what the kernel list and the developers are here for.

Jes