[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Testing test releases: do not update



On Monday 23 February 2004 12:25, Lamar Owen wrote:
> On Sunday 22 February 2004 07:17 pm, Gerry Tool wrote:
> > I heartily agree with this.  The uncoordinated mixture of rawhide
> > packages, fixed packages and original test release packages leaves a
> > very ill defined state of a system on which to base bug reports.
>
> Welcome to the real beta testing world.  IIRC, this is basically what
> things were like under the private beta test arrangement; at least that's
> how I remember the old process.

Yes, this is also how I remember it.  I believe that Lamar is absolutely 
correct.  This process is now called testing rather than "beta".  Even "beta" 
was not completely accurate since the early private beta was more like 
"alpha" testing and the public beta was the real "beta".

I define "alpha testing" as testing by "friendly users" interested in the 
ultimate target and willing to put up with lots of arrows in the back.  "Beta 
testing" is testing by users who are not necessarily friendly but the product 
is in much better shape that the "alpha" period.


>
> > Feeling impotent to help because my test sytem is hopelessly borked by
> > rawhide updates.
>
> During previous cycles I had similar issues, and did the
> 'lather/rinse/repeat' cycle numerous times on my test box.  Welcome to Beta
> Testing.  This is what it is like; if you can't handle that, wait on test2,
> 3, or 4.  Test1 is always rough around the edges.
>
> One previous cycle saw me installing from scratch three dozen times on that
> particular test box.
>
> Can it be done better?  Maybe.  Should it be done 'better'?  I don't think
> it should be, since Rawhide is just exactly that: RAW.  And, as beta
> testers, we have signed up (in my case at least) with that knowledge.  The
> test releases have big warnings about that.  In early tests, the developers
> can and will (and should not be prevented from) making large changes to
> packages; it does get better as the test number increases.
>
> People wanted 'more open'; welcome to 'more open'.

Yes, more open!  Maybe something needs to be added to the description of the 
Fedora Core testing process which defines that early snapshots will be a bit 
rough and problems can be expected.  If you want something more polished, 
wait for the later snapshots (or the final release).

As Bill Nottingham points out in one of these messages, the use of development 
for updates is intentional and reflects a continuous process of finding and 
fixing problems.  The objective is not to have a snapshot which is good but 
to have an final release that is as good as can be made.

-- 
Gene




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]