Re: ftp.rpm.org: rpm-3.0.5 packages need updated?

[Jeff Johnson <jbj JBJ ORG>]

On Mon, Aug 14, 2000 at 11:53:23AM -0400, Jim Knoble wrote:
> Circa 2000-Aug-14 10:29:09 -0400 dixit Jeff Johnson:
> 
> : >   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
> : >   -rw-r--r--   1 2369     300       1235046 Jul 21 19:14 rpm-3.0.5-7.6x.i386.rpm
> : 
> : Thes are the released version of rpm-3.0.5.
> 
> : >   ftp://ftp.redhat.com/%2Fpub/redhat/updates/6.2
> : >   -rw-rw-r--   2 root     235       1236810 Aug  9 15:53 i386/rpm-3.0.5-9.6x.i386.rpm
> : 
> : This is the errata release for all Red Hat distros.
> 
> So Red Hat is no longer using the regular release of RPM, but is
> rolling its own?
> 

( Presumably this is a request for information not just random kvetching )

Installed in our 6.x build system right now is rpm-3.0.5-10.6x.

> : IIRC, there was a problem with the value of %_sysconfdir being /usr/etc,
> : probably more.
> 
> Why aren't these changes mentioned anywhere?  Someone has to know what
> the changes are.  We're all busy folks, and it's the responsibility of
> the person who knows to transmit the knowledge.
> 

Have you even looked at %changelog? Where *else* do you wish the information
mentioned?

> CVS is a useful tool.  When i want to find out what has changed in the
> CVS repository, i should use CVS.  But i shouldn't have to use CVS to
> figure out what the differences between released packages are.  That's
> crap.  It's a symptom of shoddy, half-ass software development
> practices, and there's no excuse for it from Red Hat.  (Or from anyone
> else, for that matter).
> 

I suggested CVS becuase I believe that would give you the most complete
information with the least amount of effort. You're the 1st person I've
ever heard complain that use of CVS leads to "shoddy, half-ass software
development practices." <shrug>

> : There is yet another change to rpm-3.0.5 in CVS, to permit building
> : packages without rpmlib(VersionedDependencies). Just because a piece of
> : software is "released" doesn't mean that the software stops changing,
> : nor does it mean that every change will result in a new package being
> : QA'ed, signed, and copied to all the usual places.
> 
> To wit: "Just because software is changed doesn't mean we need to
> change its version number to distinguish it from other 'released'
> versions.  It's mostly the same, so why confuse people?"
> 
> I don't particularly expect that each change should effect the release
> of a new RPM tarball.  What i do expect is that something that is
> released is not called the same thing that was released previously and
> is not the same.  This expectation is not something i just made up.
> Lots of folks have it.  In fact, there's a rather long tradition of
> using version numbers to distinguish between ... uh, versions.
> 

And, when dealing with a "releqased" version, sometimes release is
important too. Or do you wish to see
	rpm-3.0.5.0.1a-6.7x
sort of versions. Truly, what are you asking for here?

> I don't care what changes in CVS.  I expect the CVS repository to be
> volatile; that's its function.  When packages are QAed, signed, and
> released, i *do* expect them to be copied to "all" the usual places.
> Not just one.  Especially when the one they're not in is nominally the
> source of the latest low-bugcount release of RPM.
> 

And what do you do with "must-have" fixes to "released" software? That's
what has been checked in.

73 de Jeff

-- 
Jeff Johnson	ARS N3NPQ
jbj@jbj.org	(jbj@redhat.com)
Chapel Hill, NC