PROPOSAL: Fedora user survey

[Al Dunsmuir]

Hello Michael,

Tuesday, March 9, 2010, 1:23:59 PM, you wrote:

>> On Tue, 9 Mar 2010 13:08:48 -0500, Al wrote:
>> I  want more updates. I want them to be more frequent, incremental and
>> each  reasonably  well tested. Trying to do too many changes at a time
>> not  only  leads to an increased likelihood of error, it makes it much
>> harder   to   determine   which   update  caused  a  particular  issue
>> (regression, or simple behaviour change).

> Then you need to evaluate the software at a lower level, though, instead
> of waiting for official releases. You get incremental changes only if you
> examine snapshots of the source code as found in a project's vcs.
> Upstream next official release may contain too many changes already.
> Even minor releases break badly sometimes, if a developer decided to
> rewrite code sections.

My  point is that too many developers fall into the trap of adding too
many  changes  (fixes,  or enhancements) within a given set of changes
(been there, done that, learned the hard lessions in multiple x00 KLOC
projects).  The  route to sanity is to ensure that each set of changes
is  tested  adequately (by the developer, and in real world conditions
by the end users) before moving on to the next set of changes.

The  trap that many of the "release it every 6 months" folks fall into
is the illusion that somehow this contains the damage. Often, it means
that Fedora ends up staggering under heavy impacts as enormous changes
are  periodically  (at release boundaries) thrown over the wall to the
end users, like heavy rocks launched by trebuchet.

>> I  want a Fedora playground that is up-to-date (not quite rawhide, but
>> supported  if  I  find  an issue). I am willing to accept a reasonable
>> amount  of  risk,  churn  and  extra  effort  as  part  of the cost of
>> receiving those extra updates. The primary benefit to me is seeing new
>> features and bug fixes in a useful timeframe.

> There are packagers, who won't like to take such a risk in released
> versions of Fedora, however. I would oppose also a policy that forced me to
> upgrade to latest releases without a technical requirement/rationale.

Their choice. They should know their codebase, and userbase. They have
to  measure  the  risks  in real time, realistically.

IMO,  holding back a change too long can have a large negative impact.
Making  the  change  under controlled conditions (at least some active
users  testing  and blessing each change with karma) might have better
results.
I  would  point  out  that  many  fedora users only work with released
versions  of  Fedora.  I  would hope the latest release gets some love
from  the  developers for a reasonable period (4-5 months) rather than
an  instantaneous  switch  of focus to the development release just as
the users are coming online.

One  can  argue  that  older  releases  should  get fewer changes, but
perhaps as the KDE folks have argued that means the difference between
some changes or no change.

In the end, it is their package, their choices, effort and the results
will judged by their end-users - ungrateful lot that we might be.

BTW, I'm a Gnome user, but support individual choice,
Al


-- 
Best regards,
 Al                            mailto:al.dunsmuir at sympatico.ca