Stable Release Updates types proposal (was Re: Fedora Board Meeting Recap 2010-03-11)

Jon Masters jonathan at jonmasters.org
Sat Mar 13 14:27:00 UTC 2010


On Fri, 2010-03-12 at 19:56 -0600, Matthew Woehlke wrote:
> Jon Masters wrote:

> > And prove your point that users are desperate for intrusive
> > rolling updates and won't just use Rawhide instead if they want to get
> > the very latest and greatest unbaked stuff.
> 
> First off: I'm not asking for unbaked stuff.

If you can undermine months of testing by shoving things that have had
minimal soak time into a stable release, then you are asking for unbaked
stuff. It's really just that I believe no one person is capable of
replacing the collective efforts of weeks or months of testing.

> Fourth I submit that your question is flawed and that we don't 
> currently, *generally* have "intrusive rolling updates". I have yet to 
> be hit by an update I consider "intrusive". (Even the /few/ bugs I can 
> think of that came from updates I wouldn't classify as "intrusive").

An update is intrusive if it changes expected and documented behavior
users are relying on in that release. This can be anything from plugging
your laptop into a projector and finding you can no longer display video
onto a screen, to having menu items move around without warning. I have
this crazy notion that what my laptop does today, it should do tomorrow,
exactly as it did today, unless I choose to upgrade the entire release.

How can you possibly know, when you ship some "innocuous" giant update,
that there isn't some regression introduced in there somewhere that
really annoys existing users of the software? Answer: you can't. Even
very large QA teams can miss things, and Fedora doesn't have anything
like the kind of resources to offer that kind of testing. You can't test
every situation that the software is being used under, can't know that
laptop user X now has no ability to do whatever they did yesterday. It
works for *you*, and you are not intentionally malicious, but that is of
little comfort when updates go breaking things that were working fine.

So, rather than say "don't change anything", for the love of all things
good, please just keep the intrusive massive churn to rawhide. Let users
have the new wonderful unbaked features in 6 months when they've been
tested by experienced developers running rawhide systems, and when the
user has set aside time specifically to upgrade to the latest software.

> > And while you're at it, why not tell us why people who want to update
> > constantly aren't just running Gentoo or some other compile from source
> > distribution?
> 
> Because having to compile all your own stuff constantly is a colossal pain?

Dealing with the number of Fedora updates getting shoved out to
unsuspecting users is a bigger pain. I don't even bother to update my
system daily now because I know I may need to schedule some time to fix
something when I do upgrade. It feels worse than rawhide because it's
expected to be "stable". This is not what any user wants to have. I
don't need to conduct extensive surveys to understand that no user is
desperate to have the number of updates that are going out these days.

Jon.




More information about the devel mailing list