On 18.06.25 19:53, Andres Freund wrote:
1) We don't remove support for OS versions unless they block something
Maybe it's worth clarifying the interpretation of this.
For example, for the purpose of this thread, I wouldn't consider RHEL8
to be blocking anything at the moment. It's technically blocking moving
the meson requirement past some version, but that in turn isn't really
blocking anything. You can always find a new feature in any build
dependency that you might want to use but don't really have to.
(Also, they sometimes ship updated python or openssl versions, for
example, so there is also a possible difference between support in its
original version, support with updates, and no support at all.)
But as another example, if by some miraculous development we decided to
drop autoconf for PG19, then RHEL8 would block that, but then (by some
of the policies a-e) we could drop RHEL8 support.
But my intuition is that if we did that right now, many vendors would
just have to patch the support back in, which could be great consulting
money, but wasteful overall. So if the age cutoff landed on today, it
might be too early. (Or I suppose by your rule #4 they could just keep
supplying upstream patches to keep the support alive.)
(I initially thought that RHEL8 was a typo for RHEL7, because we still
support RHEL7!! We should drop that first!)
