commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From robert burrell donkin <>
Subject Re: Second patch, somewhat easier, replacing the first one....
Date Tue, 18 Dec 2001 18:45:06 GMT

On Tuesday, December 18, 2001, at 03:37 AM, Craig R. McClanahan wrote:

> On Tue, 18 Dec 2001, Mika Goeckel wrote:
>> Date: Tue, 18 Dec 2001 00:09:48 +0100
>> From: Mika Goeckel <>
>> Reply-To: Jakarta Commons Developers List <commons-
>> To: Jakarta Commons Developers List <>
>> Subject: Re: Second patch, somewhat easier, replacing the first  one....
>> Rob,
>> do you think of a callback in the digester to get the information what 
>> last
>> matched?
>> The problem with that is that the digester would need to store the
>> information in an instance variable then.
>> Ok, digester is not multithread-safe, but my blood always runs cold when 
>> I
>> think about 'global' variables.
>> I'd prefer a clean way of passing the information to the configuration
>> method.
> My blood shivers at things like that as well, but there's already 25 or so
> other instance variables ... one more wouldn't matter.  I can't really see
> why anyone would need a thread-safe Digester, so I don't see much point in
> making it so (or paying the extra synchronization costs).

yes, i agree, i know it feels wrong. unfortunately calling digester back 
is IMHO the lesser of two evils.

when we were thinking about those other rules, we considered breaking or 
extending the rule interface instead. the problem is that once you start 
modifying the rule interface to pass more information in, when do you stop?
  with two parameters, ten, twenty? every addition adds complexity or 
breaks compatibility.

hopefully sometime in the future we might know pretty well enough what's 
required and then we could think about making it work differently.

- robert

To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message