Description of problem: In the past many config files for bind have been part of the rpm package caching-nameserver. This has been changed to be now "bind-config" as sub-package of the bind src.rpm. I think we should only change the naming if we really have to, but otherwise stay compatibel. E.g. "caching-nameserver" is part of our RHEL documentation and many other docus. Can we re-consider keeping to the old naming scheme? Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Yes, I'll rename the bind-config sub-package, that replaces caching-nameserver, back to caching-nameserver for RHEL-5, to be in-line with our RHEL documentation; but now Fedora users are used to bind-config, I think we should keep the name the same for Fedora. Moving this bug to RHEL-5 (or trying to...)
Wouldn't it be way easier to stay in sync between RHEL and FC? regards, Florian La Roche
I'm going to turn it back, but we need to restore the caching-nameserver package for FC6 (and for RHEL5?) first.
The bind-config subpackage has a Provides: caching-nameserver, so automated tools, such as yum (and up2date?) when prompted for: install caching-nameserver will get the right tool. Changing this yet again is getting a little silly, but I'll go along with it if I have to. However I would like to point out that in the past, caching-nameserver was a standalone package, its own srpm. Then bind-config was created as a subpackage of bind. Could we not just change the name of bind-config subpackage to caching-nameserver (with appropriate Obsoletes/Provides bind-config)? This keeps it all in the bind package and there would be no need for "adding" a package to the collections.
(In reply to comment #4) > However I would like to point out that in the past, caching-nameserver was a > standalone package, its own srpm. Then bind-config was created as a subpackage > of bind. Could we not just change the name of bind-config subpackage to > caching-nameserver (with appropriate Obsoletes/Provides bind-config)? This > keeps it all in the bind package and there would be no need for "adding" a > package to the collections. Renamimg bind-config to caching-nameserver and leave it as a subpackage sounds good to me...
(In reply to comment #4) > Could we not just change the name of bind-config subpackage to > caching-nameserver (with appropriate Obsoletes/Provides bind-config)? This > keeps it all in the bind package and there would be no need for "adding" a > package to the collections. The updated bind-9.3.2-35 is here, could you please move it in RHEL-5?
So, can anybody here mark this bug and package (bind-9.3.2-36.fc6) as a Beta blocker? -------- Original Message -------- Subject: Re: rpms/bind/devel bind.spec,1.118,1.119 Date: Mon, 14 Aug 2006 08:11:29 -0400 From: Jesse Keating <jkeating> To: Martin Stransky <stransky> On Monday 14 August 2006 07:53, Martin Stransky wrote: > So, can I close the #197493? It would need to be a beta blocker package and if it is, you'd have to request that it get pulled in to the RHEL5 Beta through 'release-engineering' using "RHEL5B1 INCLUDE: bind-9.3.2-36.fc6" subject line. Please be sure it is accepted as a blocker bug before you ask for inclusion. -- Jesse Keating Release Engineer: Fedora
I think this should be important enough to go into beta. Otherwise we start an Obsoletes: loop where caching-nameserver obsoletes bind-config and vica versa. (That is probably not handeled by our update tools, so should really be avoided.) Proposing as beta-blocker. New packages are already in FC6 trees AFAIK. regards, Florian La Roche
Indeed, the package names need to be nailed down prior to B1. If we're planning on going out in RHEL5 GA with caching-nameserver being a subpackage of the bind SRPM, then let's make that change now and get it into the B1 trees.
comps file then also needs adjustments. regards, Florian La Roche
I think we're okay as far as comps is concerned -- the comps files for the various products don't mention the package by either name.
I can not see any substential impact in one or the other way. We might have documentation mentioning it, but that can be corrected. Customers might rely on it. On the other hand we do more intrusive changes, so probably they check before expecting something like this not to change in a major release. I am OK with either way. Currently the package in Rawhide is called caching-nameserver, it is not at all in Comps but gets pulled in as "caching-nameserver" by NetworkManager. So probably leave it as it is (means "caching-nameserver")
fixed in CVS, built in bind-9.3.2-37.fc6 (with caching-nameserver as a bind subpackage)
I've raised a request to release engineering to move bind-9.3.2-37.fc6 into the rhel-5 beta tree: https://engineering.redhat.com/rt3/Ticket/Display.html?id=3777
Done.
$ pwd /mnt/redhat/rel-eng/RHEL5-Server-20060825.1/4.91 $ find . -name "caching-nameserver*" -print ./i386/os/Server/caching-nameserver-9.3.2-37.fc6.i386.rpm ./ppc/os/Server/caching-nameserver-9.3.2-37.fc6.ppc.rpm ./s390x/os/Server/caching-nameserver-9.3.2-37.fc6.s390x.rpm ./x86_64/os/Server/caching-nameserver-9.3.2-37.fc6.x86_64.rpm ./ia64/os/Server/caching-nameserver-9.3.2-37.fc6.ia64.rpm