Update/install experience

[Ralf Corsepius]

On 12/16/2009 12:48 AM, Kevin Fenzi wrote:
> On Tue, 15 Dec 2009 11:06:33 -0500
> Will Woods<wwoods at redhat.com>  wrote:
>
> ...snip...
>
>>> So, say we have 30updates/day currently.
>>
>> The average starts at around 25-30/day for the first 4 months and then
>> declines to 20 or less over the life of the release. Check the Bodhi
>> "Metrics" pages for some data.
>
> Yep.
>
>>> With this model they would pile up and then be tested in a unit?
>>> Then once a week all 210 of them would be pushed out?
>>> (The once a day blorp becomes a larger once a week blorp).
>>
>> Me, I'd prefer a once-a-month drop of changes, but yes. Same amount of
>> changes, just batched into larger, periodic system updates.

To me this sounds like bad idea - Primarily for 3 reasons:

a) This contradicts Fedora's aims (fast pace) and is non-helpful to 
everybody. developers and package maintainers should be interested in 
seeing their bugs found and fixed ASAP, as well are users interested in 
seeing bugs they are fixed ASAP.

This is want makes the key difference between Fedora and rawhide 
(volatile and unstable), RHEL (stagnation) and other distros with long 
bugfixing cycles.


b) This strategy is non-helpful to users, because "small haps" of 
updates are "digestable", but "big chunks" of updates will "choke" them 
once a month.

It will lead to users to opt to delay updating at times when it is 
"convenient" to them => they will expose themselves to broken packages 
and to security risk much longer

It will contribute to maintainers not being able to fix bugs => It leads 
to users struggling with bugs == a regression of "user experience".


c) This leads to a worser quality of Fedora, because unlike to what you 
believe, the mass of testing is performed by users, when using Fedora, 
not by RH's testing group nor in "*-testing".


>> The destabilizing effect is much, much less than (e.g.) the daily
>> changes we get during the freezes for Alpha/Beta. It's a manageable
>> amount.
IMO, this is a wild claim without any justification - I question it.

It will lead to more "slience" and "less churn", but not to a better 
quality of the distro nor to a better "user experience".

Ralf