Re: OSABI

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Mon, 22 Aug 2016 16:39:56 +0200
    From:        Manuel Bouyer <bouyer%antioche.eu.org@localhost>
    Message-ID:  <20160822143956.GA2693%asim.lip6.fr@localhost>

  | this is broken then, as it tries to identify both the kernel and
  | userland ABI.

Agreed.   Same reasons.    Further, other than for kmem grovellors
(which should really all be fixed to use some more stable mechanism -
most have been already) and (rarely) *ioctl() interfaces, the kernel
version matters not at all - what matters is what libc gives access to,
and that's controlled by which userland is installed, which has nothing
at all to do with what uname(2) returns.

The problem we have is that while it is possible to parse /etc/release,
we really have no rational way to determine the userland version or
capabilities - at either runtime or compile time.

kre

ps: I know that the kernel version is constrained by what libc, and
others, expects.