commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David Gilliland" <dgillilan...@users.sourceforge.net>
Subject Re: Proposal for the Commons Sandbox
Date Thu, 18 Mar 2004 15:31:52 GMT
>
>>It could form a separate commons project. It could provide
>>alternative implementations/plugins to the [convert] project.
>
Plz forgive my delay in following up on this (things got really crazy at work).  Anyway, thanks
for the suggestion about [convert]; I really like that approach to conversion in general,
and have incorporated support for it into Jestr.  As of the beta7 release, the jestr.properties
file now supports syntax allowing the user to associate any Converter or ConvertRegistry with
any arbitrary portion of the object graph.  Thus the user can leverage the stringifier selection
infrastructure of Jestr (predicates, priorities, etc.) combined with the libraries of converters
that may be expected to build on the Converter interface.

I also included a class that implements the Converter interface by delegating to Jestr.  Thus
Jestr can be treated as "just another to-string converter", which is a nice approach because
it insulates the user from dependencies on any particular stringifier implementation, similar
to the way [logging] insulates the user from dependencies on any particular logger.

See this example for details:
http://jestr.sourceforge.net/jestr/examples/example6/package-summary.html

As to the question of whether Jestr's feature set should be incorporated into [convert], IMHO
this would not be a good idea, for reasons similar to those you list for Jestr being a mismatch
in [lang].  For one thing, Jestr is about 5x the size of [convert].  Plus it appears to be
contrary to the philosphy of [convert], which seems oriented toward providing (a) interfaces
defining what a converter is and (b) basic implementations of converters for fundamental types.
 Jestr goes *VERY* far beyond this in the particular direction of to-string conversion, and
is unconcerned with all other kinds of conversion.



---------- Original Message ----------------------------------
From: "David Gilliland" <dgilliland62@users.sourceforge.net>
Reply-To: <dgilliland62@users.sourceforge.net>
Date:  Wed, 10 Mar 2004 11:06:58 -0500

>>jestr breaks the goals of lang on grounds of:
>>- size
>>- dependencies
>>- potential to grow
>>- 'extensible'
>All extremely good points.  Just speaking on the matter of size--Jestr would increase
the line count of [lang] by something like 50% and more than triple the number of classes/interfaces.
>
>>It could form a separate commons project. It could provide
>>alternative implementations/plugins to the [convert] project.
>A separate project was my original suggestion.  I'm not too familiar with [convert] but
will investigate that.  Thanks.
>
>--David
>
>




---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org


Mime
View raw message