[syslinux] installing latest rpm on RHEL
R P Herrold
herrold at owlriver.com
Wed Apr 7 11:24:15 PDT 2010
On Wed, 7 Apr 2010, Dag Wieers wrote:
> This is a very annoying problem with a change in the RPM format (without
> providing any backported support in newer RHEL releases or compatibility).
>
> Not good for something that is a part of the LSB standard :-/
Hi, dag
That was my initial reaction to the change in the native
compression method used by teh lastest Fedora's and 'RawHide'
rpmbuild, and before that to the move away from --md5sums --
but the problem is that the rpmbuild in the later Fedoras's
(and no doubt to move to RHEL at the 6 level) is that they are
WRITING to a format (and not LZMA{2|}) that 'eveyone else' is
This is Red Hat's RPM maintainer moving its (and frankly the
Enterprise RPM branch of the RPM-scism with market share)
featureset in a fashion to force a person to run at least one
instance with 'the latest and greatest' to be able to portibly
MOVE SRPMs around. A former maintainer at Red Hat followed a
different course that bent over backward to maintain
backwarerd compatability, but times change ...
There is not a LSB aspect to the matter which standard merely
requires being ablte to READ and install content packaged in
the RPM wrapper and without triggers. Nothing in the LSB
requires creating RPMs using rpmbuild, or any ogther tool
[indeed, the LSB folks build a per based package builder in
the last couple of years]
heh: ... unless one wants to CREATE an LSB conformant RPM
under Fedora, and one assumes, soon, RHEL 6) without figuring
out the compatability macro changes.
We have time for that and I'll post a blog post on the matter
once the dust settles with a 'recipe'
One has to think this is intentional marginalizing of the
rpm-build part of rpm, and a move to building packages
Pythonically -- and if one thinks this through, it is just Red
Hat deciding it is time to squeeze this 'choke point' to push
its commercial Enterprise competitor and its buildservice into
a smaller and smaller area of relevance
-- Russ herrold
More information about the Syslinux
mailing list