On Jan 5, 2009, at 1:30 PM, Michael Keith wrote:
I don't understand why literal names are more meaningless than
abstract ones.
They tend to be overused because we have so many concepts to name, and
each overloading of the name makes it less meaningful.
For example, "context" is one of the most overused terms on the server
side: transaction context, security context, class loader context,
JNDI naming context, language context (scope), etc. Calling something
a "contextual service" doesn't distinguish it from any other kind of
service, because "context" has a low intrinsic information content.
For example, I find something like Contextual Services more
meaningful than "web beans".
Since "web beans" isn't a great name either, that's not fair. :-)
"weave" has kind of already been claimed by the AOP
community to
mean byte-enhance. I
expect that by and large that would be the connotation with anything
named after it.
The spec does have a bunch of byte-code enhancement and AOP stuff. :-)
-- Scott
> -----Original Message-----
> From: Scott Ferguson [mailto:ferg@caucho.com]
> Sent: Monday, January 05, 2009 1:06 PM
> To: Michael Keith
> Cc: Gavin King; Java Community Process JSR #299 Expert List;
> Matt Drees;
> Jim Knutson; WebBeans
> Subject: Re: New name
>
>
>
> On Jan 5, 2009, at 5:39 AM, Michael Keith wrote:
>
>>
>> One of the problems with the existing name is that it names a new
>> type of object,
>> which gives the impression that a new "component" is being
> introduced.
>> Rather, this spec is supposed to be introducing a new set of
>> container services, so
>> a better direction might be to name it around the sevices
> that it is
>> offering and
>> not the objects that are the beneficiaries of those services (and
>> are supposed
>> to already exist outside of this spec).
>
> Excellent point.
>
>> A few ideas, just to illustrate what I mean, and start the naming
>> juices flowing
>> in this direction:
>>
>> Context and Injection Services
>> Container Object Services
>> Container Contexts and Injection
>> Contextual Support for Container Objects
>
> I'm not sure I like the literal names. Since the problem to
> be solved
> is so general, literal names are also abstract and somewhat
> meaningless. For example, "container", "context" and
"object" are
> used everywhere, so they don't help explain how this spec is
> different.
>
>> The other option is to give it an arbitrary name and let the
>> contents speak
>> for itself (a la "Swing", and other similar randomly-named
>> technologies).
>> "Fred" has a nice ring to it ;-)
>
> I like "weaver" (or "weave", etc.) It fits the problem because
a
> weaver pulls together materials (components/wool) using a
> pattern/plan
> (config or rug pattern), creating the final completed product
> (application or rug). And it has a vivid image, so you can remember
> it and distinguish it from other specs. The "weaver pattern" could
> even be a replacement for "IoC/DI", a name no one really likes.
>
> -- Scott
>
>