problem with readers.conf

[Bill Davidsen]

Jeffrey M. Vinocur wrote:
> On Sun, 18 Nov 2007, Javi Pardo wrote:
>
>   
>> First, Im sorry for my bad english
>>     
>
> That's all right, we'll do our best.
>
>
>   
>> I ham in this moment INN2.2.3 with public access and varius privates 
>> access (for diferents groups)
>> In this moment i have other experimental server with INN 2.4.3 for 
>> migrate the actual INN server
>>     
>
>   
>> I have in this moment implemented this reader.conf (posted in the final 
>> of this post)  but not work... 
>>     
>
> I see several problems with the readers.conf you posted.  I think the 
> concept you're missing is about how auth blocks work.  It's like this:
>
>   - a connection comes in
>   - INN looks at the IP address and compares it to host: lines
>   - the last matching auth block is identified
>   - only that block is used -- INN does not continue on to others
>   

That would be clearer without the last half of that sentence. INN does 
continue to evaluate all auth blocks and only actually uses the last 
matching block. This behavior can lead to a *lot* of auth blocks to get 
the desired behavior. In retrospect it would have been nice to have an 
option to do auth as either first match or last match to fit the site 
requirements with least complexity.
> So probably you need fewer auth blocks with more complexity in each one.
>
> It's hard to give you specific advice because I'm not sure exactly what 
> you're trying to accomplish.  Do you have an old nnrp.access that gives 
> the desired behavior?  That might help us figure it out.  Otherwise, can 
> you describe in detail what you intend?
>
>
>   


-- 
bill davidsen <davidsen at tmr.com>
  CTO TMR Associates, Inc
  Doing interesting things with small computers since 1979