Effect of 0.17.3 lexer change

[David Relson]

On Fri, 19 Mar 2004 13:35:21 +0100
Boris 'pi' Piwinger wrote:

> David Relson wrote:
> 
> >> > Bogofilter-0.17.3, a.k.a. "Code Clean-up Release - Phase 2", has
> >> > been released.
> >> 
> >> I did a check on my mail collection about the lexer change.
> >> Of 11529 ham messages 1 is now rated incorrectly.
> >> Of 14397 spam messages 4 are now rated incorrectly.
> >> 
> >> So to my surprise the effect is very small. YMMV, in
> >> particular if you use the block_on_subnets option.
> > 
> > I assume you're referring to the IP Address fix ???
> 
> Yes, has there been anything else?

No.  Having forgotten about making that change, I had to search the
ChangeLog to figure out what you were referring to.  Guess I've got too
many different things on my mind ;-)

> > I'm presently experimenting with another lexer change.  A while back
> > I noticed that mime header parts ( the Content-Type,
> > Content-Disposition, etc lines after a mime boundary) are tagged as
> > head:Content-Type, etc and thought it wrong.   I'm presently using
> > "mime:" as the tag for these lines which distinguishes them from the
> > same lines in the message's header.  If you're interested in
> > experimenting/testing I've attached a patch.  I'm thinking of
> > including this change in 0.17.4 :-)
> 
> I am not really sure if it is all that wrong as it is now.
> It might be useful to tag those lines by their name, like:
> content-type: text
> content-type: html
> content-type: charset
> content-type: utf8
> etc.
> 
> But maybe mime also does the job.

Likely using "type:" and "disp:" would do even better.  I'd classify
that more as an enhancement than a fix.