cbolin
25th November 2004, 10:15 PM
One of the things I like about Fedora is the speed of releases - undoubtedly there is a quality tradeoff in this velocity.
However FC3 updates seem to not have been tested at all before release.
I installed FC3 the day after release. All was fine - happy other than /dev/cdrom was gone, but easily fixed.
One day later I ran up2date and all of a sudden my soundcard stopped working. For whatever reason gnome-volume-control changed the analog input to off.
A week later I ran up2date and got an update to X which apparently conflicts with the installed version of eggcups (not prelinked?) which causes a hard crash.
Luckily there are smart people that have time to figure out each of these failures post release - but I have to wonder, it is pretty obvious no one even tested - and in the above cases meaning simply ran - the applications after a prereleased update.
Doesn't it make sense to have a BVT or smoketest that at least runs each included module and looks for failures? and to distribute to a test bed sample of at least the most common hardware/software configs before going public?
I like fedora, I'd just like to have updates that don't require research and troubleshooting to have to apply. Given updates are available so often - a happy medium between release velocity and quality needs to be met?
However FC3 updates seem to not have been tested at all before release.
I installed FC3 the day after release. All was fine - happy other than /dev/cdrom was gone, but easily fixed.
One day later I ran up2date and all of a sudden my soundcard stopped working. For whatever reason gnome-volume-control changed the analog input to off.
A week later I ran up2date and got an update to X which apparently conflicts with the installed version of eggcups (not prelinked?) which causes a hard crash.
Luckily there are smart people that have time to figure out each of these failures post release - but I have to wonder, it is pretty obvious no one even tested - and in the above cases meaning simply ran - the applications after a prereleased update.
Doesn't it make sense to have a BVT or smoketest that at least runs each included module and looks for failures? and to distribute to a test bed sample of at least the most common hardware/software configs before going public?
I like fedora, I'd just like to have updates that don't require research and troubleshooting to have to apply. Given updates are available so often - a happy medium between release velocity and quality needs to be met?
