On Mon, Apr 13, 1998 at 12:28:15AM -0500, Manoj Srivastava wrote: > > Congratulations! You have just introduced a subtle bug on your > system. It may work, and possibly never cause a problem, but > there is a bomb ticking away, waiting to explode ;-)
Which bug is that? If it's really that big of a deal I was considering rebuilding glibc anyway with egcs and adding to it the dependancy of either egcc or gcc. This would satisfy both issues? > There is a reason there is a versioned dependency for > libc6-dev. The reasons are explained in a libc6-dev FAQ. I have also > posted it in a related document. > > I think I have changed my mind. I think libc6 should really > get a package all its own, called libc6-kernel-headers. I do not know > whether I can push it into 2.0, but I shall try. Please do. I had not realized I had created a problem by doing this. I shall correct it directly. > All this silly snipping of links and upgrading to incompatible > headers may cease then. mmm, no. As long as things like OSS/Linux demand that your kernel headers in /usr/include/linux be the same as those of your current kernel before their pathetic install script will actually install, people will likely do what I did, oblivious to the potential consiquences. A question which comes to my curious mind... is there a way a program running as root can ask the kernel things like "do you support modules and module versioning?" or is the above script which hung my machine without so much as an oops from 2.1.82 till 2.1.89 the only way an installer can check these things? (it reads autoconf.h) (BTW, I am -GLAD- I yanked the CS4232 card and put the ES1688 in--no more ISA PnP and I can now compile OSS/Free which never worked with the CS4232)
pgppjmHSynyMe.pgp
Description: PGP signature
