Mime-Version: 1.0
Message-Id: <p04330100b64ab17bed63@[193.112.141.252]>
Date: Wed, 29 Nov 2000 13:21:19 +0000
To: Perforce Defect Tracking Integration Project staff <p4dti-staff@ravenbrook.com>
From: Richard Brooksby <rb@ravenbrook.com>
Subject: Setting items to be replicated
Content-Type: text/plain; charset="us-ascii" ; format="flowed"

This is a record of a design meeting between RB and GDR and NB about 
controlling the set of issues that are replicated.

The replicator should only replicate things that have been created or 
modified after the first time that it is run, and not since the 
beginning of time.  Some organizations have very large existing 
databases, and don't want all of their historical data to appear in 
Perforce.  (Quokka Sports was an example.)

However, some organizations will want to replicate older issues.  We 
should provide a way to do it.

1. We could set the replicator's "last change replicated" time back 
so that it thinks that things have changed and replicates older 
issues.  However, this will also cause it to examine a potentially 
large set of changes twice, and doesn't give very good control.

2. We could provide a command which causes a set of issues to be 
replicated.  Once replicated, things stay replicated, so that would 
be enough to get them going.  We could then provide a number of ways 
of specifying which issues are needed: by project, by date range, by 
issue identifier, and so on.

Method (2) is more general and flexible, and depends less on the 
internal workings of the replicator.  I think we should do it, but 
not for the beta.