Yeah, but that's not what was said.
What he said was that
(a) he thought he had good reasons for doing so, but that
(b) it was a mistake.
Unfortunately, (a) almost completely overshadowed (b).
I've never heard (b). I've heard him claim that the original addition
of llseek() was a mistake, and should have never been added in libc5 or
glibc2. This belies the fact that llseek() existed under Solaris and
AIX, and predates the LFS API standard.
Ulrich's claim was that original addition of llseek() in glibc was a
mistake (which I think is wrong, since llseek() is a historical
interface which has been around for a very long time), and to assure
backwards compatibility with the few people using glibc, he kept the
llseek() function in the library while deleting the prototype, this
screwing over the large number of people who hadn't yet converted over
from libc5 to glibc2, and who had > 2GB filesystems (and who were
compiling their own utilities and didn't know about this compatibility
trap which had been laid out for them).
- Ted
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu