Trackin' stable vs. releases

[Rick Moen]

A listmember who "doesn't want to join the debate" wrote me off list.
His identity is accordingly snipped, but I really didn't feel like
starting discussions in private mail on this matter.

I trust and hope that I won't offend him by posting quoted text 
without attribution.  If I'm wrong, <shrug>.  Sorry.

> I haven't been paying close attention to this one, but I suspect that
> the line of thinking goes like this:
> 
> You start at stable, but then find that some of the packages you want
> are only on testing or unstable, so, you investigate and find it's
> working fairly well, and IS more or less stable as long as you check
> for major new bugs before you update your packages.  Now, this implies
> that "unstable" may be ALMOST to the stability level of testing.  Then
> a release comes.  What happens is that unstable suddenly becomes VERY
> unstable.  The version you really wanted has become testing, and it
> gets a bit tricky.

But it's not really true that it's "the next version" that just became testing,
at that point:  Yesterday, testing was populated by packages from
unstable that met quartantine criteria.[1]  Today, it's populated by
packages that meet those same criteria.

If you change the sources.list entry from "testing" to "woody" around
that time, what you're really saying is "I've for some reason suddenly
decided that the quarantine criteria that looked good yesterday have
begun looking lousy today."

I suppose one could arrive at that view, but it seems odd.

Working hypothesis:  A lot of people haven't thought this matter
through, very well.

[1] These:  http://people.debian.org/~jules/testingfaq.html