[PEAK] removing adaptions

[Phillip J. Eby]

At 10:08 PM 10/25/04 +0200, Gabriel Jägenstedt wrote:
>Hi!
>
>I've got a system that is meant to be adaptable to different languages.
>I've tried several different approaches and have finaly found one that I
>think fits the purpose.
>
>I only use protocols.
>
>Each language that is supported has to create two things. An Interface
>and a language module.
>
>The language module provides ILang to show that it's a language and
>IEnglish to show that it's the english language.
>
>Inside the module there resides different adapters EnglishParserAdapter
>EnglishCommandAdapter and so on.
>I have gotten everything to work perfectly with a bit of directory
>scanning, importing and adapting so that I can call lingo('english',
>IParser) and this returns a parser that understands english.
>
>Now comes the problem if one desides to support different languages
>during sessions there is already another adapter in play.
>Is there a good way to disable an adapter? Or is my design faulty as per
>usual =)

I don't know, because there's not enough here for me to even begin to 
understand what you're doing, how, or why.  :)

But I will attempt to dissect a couple of clues anyway.  First, it's not 
clear to me that you are "adapting" anything here.  You've never once 
mentioned what you're adapting *from*.

It sounds more to me like you are registering *implementations* of various 
things that you need, and then using them.  If that's so, then it's largely 
unrelated to adaptation; it's more like using PEAK's configuration 
facilities to register the availability of various services.  (And PEAK 
offers lots of very convenient ways to do things that sound a lot like the 
things you mention here.)

However, if you *are* using adaptation to adapt *from* something, then you 
should keep that original something around if you need to adapt it to 
something else, as is strongly suggested by the PyProtocols manual.