"the rpm question" - summary

[Ingo T. Storm]

I have been silent so far because I seem to have joined the list late and
did not know all pros and cons of the two sides. Unfortunately the
discussion has now rather become political. Not really knowing any of the
people involved, I'd like to get back to the issue, because I consider it
VERY important. I hereby also acknowledge that I might be an idiot on the
technical side, have no experience with rh8/9, but only rh72/73/RHEL21.

I think it is fair to summarize that the "pro rpm upgrade" logic is
technical with two objectives:

A. fix bugs in rpm
B. simplify srpm maintainance for all platforms

The people "against" have two points, too, although less technical:

C. it might break things
D. I thought FedoraLegacy is minimal invasive surgery that only patches
holes?

>From my point of view, A is valid, C is unproven. Let's put aside B for the
moment.

I think it all comes down to D, which is rather about opinion and directions
than technical fact and touches an important question: How much _Fedora.us_
is in _FedoraLegacy_? To me it looks like it's rather the "Fedora.us" core
people would like to stay on a successful track (same rpm for everyone),
whereas a lot of newcomers want as little change to their platforms and most
of them only want security errata "the way we used to get them": backports
of security or important bug fixes to the software versions the have on
their machines.

I can see the benifits of a new version of rpm that is the same across all
supported platforms. (B is definitely a big point here.) But I personally am
here for patches to the version of software package XYZ, _not_ for a new
version that patches the hole and bugs and introduces new features - and
perhaps breaks my config due to new default behaviour (anyone not thinking
"php" here?). If rpm has bugs - fine. But they don't seem to be bad enough
that a lot of people (users, not Fedora.us programmers) haven't learned to
live with them. I run servers. Some have already gone RHEL, others will, my
sparcs and alphas just can't. But I need all my servers to run reliably, and
that means "as little change as possible".

Add to that the possible complexity. Assuming all people who _want help_
also _want to help_, so their distros get in the boat, FedoraLegacy will
support at least 10 distributions/versions on at least 4 hardware platforms.
We're probably talking about 8 different glibc versions. While it might have
been viable for Fedora.us to have one version of rpm tried and tested (i.e.
it rebuilds nicely all base rpms plus all errata of all platforms), I am
afraid this cannot be true for FedoraLegacy.

I really don't know how to settle this issue, but I simply cannot wait for a
decision until mid december. If I have to upgrade rpm on sparc and alpha, I
will have to port it myself. This is too much work and probably too
difficult for me, too. Then I'd rather initiate a very much smaller project
"keep Enigma and Valhalla alive", because that's all I run. This would be
about _only_ backporting RHEL/ia32 errata to rh72 and rh73 on ia32 and then
port those to Aurora and rh72axp. Doable by a handful of people, but still a
huge waste of effort if done in parallel to FedoraLegacy.

So please folks, everybody think again from scratch. Maybe about what
FedoraLegacy really is about. Is a new version of rpm a nice-to-have or does
it "cut the work/bugs in half" or is it a clear must-have? And then let's
vote on the list - soon. For someone with rh72/73 systems out there on the
Internet, January 1st isn't really far away.

Thanks,

Ingo