Difference between a test update and a release of same

[David Timms]

Horst H. von Brand wrote:
> Rahul Sundaram <sundaram at fedoraproject.org> wrote:
>> David Timms wrote:
>>> Hi, if I am testing a particular update say dvd+rw-tools, if the test
>>> package is pushed into updates, are the rpm's identical ?
>>> Or are they identical before they are signed for each repository
>>> {updates, updates-testing} ?
>> Yes.
> 
> Expanding a bit on this, the Developer Guidelines ask for the release to be
> bumped each time a new package is built, however minor the change.
Which makes sense or you would have total confusion with v4.5.6 works 
here, but not over here because they could have been different version 
of identical package.

 From the point of view of an updates-testing compared to an updates 
package of the same version, is the package actually rebuilt (ie from 
.src/spec), or is it just the signing authority that is different ?

Actually, I found a package that is currently in both repos:
$ ls -lgG virt-manager-0.2.6-3.fc6.i386.rpm*
-rw-r--r-- 1 566479 Jan  7 14:03 virt-manager-0.2.6-3.fc6.i386.rpm
-rw-r--r-- 1 566479 Jan  7 14:03 virt-manager-0.2.6-3.fc6.i386.rpm.test
   -same size
$ md5sum vir*
ebcb47c6fba4a4a6707a80c375507984  virt-manager-0.2.6-3.fc6.i386.rpm
6f755debee94a707ae2513cb7d8d59bc  virt-manager-0.2.6-3.fc6.i386.rpm.test
   -but definitely different.
$ rpmdiff virt-manager-0.2.6-3.fc6.i386.rpm 
virt-manager-0.2.6-3.fc6.i386.rpm.test
$
   -but no changed files
$ diff -C 0 z.t.txt z.txt
*** info.test.txt  2007-01-07 14:07:23.000000000 +1100
--- info.txt       2007-01-07 14:07:30.000000000 +1100
***************
*** 7 ****
! Signature   : DSA/SHA1, Fri 22 Dec 2006 08:36:38 AM EST, Key ID 
da84cbd430c9ecf8
--- 7 ----
! Signature   : DSA/SHA1, Wed 03 Jan 2007 02:18:34 AM EST, Key ID 
b44269d04f2a6fd2

   =>only the signature date and key ID is different. build host and 
date are identical.

I wanted to understand how a production system might be affected by 
using a updates-testing package {that solves a problem}. It seems if 
that particular -testing package is eventually moved to the stable 
updates, then the only detectable difference is the signing date and 
time of the rpm, ie it can not introduce any binary/code differences 
between a system that has the package version from either repo.

Thanks Horst.

DaveT.