Re: security in testing
On Wed, May 14, 2003 at 10:14:53AM -0500, Gunnar Wolf wrote:
I'm sorry, I am on a public terminal, and can't quite remember where I
read it - But testing should always be close to a releasable state.
That assumption is both false and absurd. Testing has exactly two
advantages over unstable--1) all dependencies are satisfied and 2) known
rc bugs don't propagate to testing. In all other respects unstable is
better. (Security problems, rc bugs not noticed during the first two
weeks, etc.)
Of course, currently it is not due to debian-installer is still not
ready for human consumption... But the bulk of testing should always be
usable.
"should" based on what? when has this been true?
Testing was introduced because the inadequacy of freezing
unstable for months in order to get a release done.
And to this day testing is primarily useful near a release, when testing
*does* get security attention, and when fixes are pushed to testing. We
are not currently near a release.
Mike Stone
Reply to: