Re: WHAT is a `specific snapshot' ? (Was: New policy...)

[Marc Espie <Marc_(_dot_)_Espie_(_at_)_liafa_(_dot_)_jussieu_(_dot_)_fr>]

On Tue, Oct 26, 1999 at 12:08:58PM -0600, Theo de Raadt wrote:
> Chris Turan is trying to build a framework where packages will be reliable.

And I am offering constructive criticism.

As I said:
> > Namely, WHAT is a `specific snapshot' ? How does this translate in a tree
> > with 8 separate architectures, where snapshots NEVER occur at the same time.

> > Right now, Theo is building 5 snapshots out of 8, which is already too much.
> > If I want my amiga to survive, I need to continue building snapshots.
> > If I want to check things properly with your scheme, I need to be able
> > to reproduce a specific snapshot.

> > Namely, this needs we do neeed to tag the source tree whenever someone build
> > snapshots... (something such as SNAPSHOT-991101-AMIGA for me).
> > this is not hard to do, actually. Major projects such as gcc
> > routinely tag their cvs tree each time they produce a snapshot.


To sum this up, IF you want that scheme to be reliable, you need to make it
possible for people involved in snapshots building to get some accurate and
reproducible results. Specifically, you want to have the major architectures
provide SYNCHRONIZED snapshots, that reflect the same state of the ports tree,
otherwise you're going to run into false positives, OR dismiss actual problems
that occur on a specific architecture, thinking they are a transient problem
associated with a specific snapshot not being in synch with the rest of the
snapshots.
-- 
	Marc Espie		
|anime, sf, juggling, unicycle, acrobatics, comics...
|AmigaOS, OpenBSD, C++, perl, Icon, PostScript...
| `real programmers don't die, they just get out of beta'