"url:" counts

[David Relson]

On Fri, 9 Jan 2004 03:14:20 +0100
Matthias Andree <matthias.andree at gmx.de> wrote:

> On Thu, 08 Jan 2004, David Relson wrote:
> 
> > Prompted by Matt's comment on the misnaming of "url:" tokens, I
> > counted what's in my database and how many have very low or very
> > high scores. 
> 
> We should stuff a .BOGOFILTER_FORMAT version integer into any data
> base that doesn't have it. We'd start with setting
> .BOGOFILTER_FORMAT=1602 in 0.16.2 to use anything that doesn't look
> like the first version, and whenever parsing changes, we can prompt
> the user what to do: convert with bogoupgrade (preferred) or ignore
> (which is discouraged but would just bump the .BOGOFILTER_FORMAT
> version).
> 
> Bogofilter would only open data bases that have a matching
> .BOGOFILTER_FORMAT version and refer to bogoupgrade otherwise. After
> url: is renamed to ip:, .BOGOFILTER_FORMAT will be set to 1603.
> 
> bogoupgrade would then know to s/url:/ip:/ for upgrading from 1602 to
> 1603.
> 
> After 0.16.2, the .BOGOFILTER_FORMAT would likely decouple from the
> actual version, so we might consider starting with 21, 23 or 42
> instead of 1602 (any relation with existing books or living or dead
> authors is purely intentional).

Yes.  We could do that.  My first reaction is that it's not really
needed -- especially because we can live with "url:".  Changing to "ip:"
is a nicety, not a necessity.

Anyhow, I don't want to jump into version numbers, without due
consideration.  I want to know exactly what we're doing and why we're
doing it.  I've had bad experiences with file version numbers in the
past and want a good plan before doing anything.

Were I to pick a version number, I'd likely start at 41, to honor the
first version of year 2004.

David