Re: [PROPOSAL] lifecycle release
Stefano Mazzocchi <[email protected]> Tue, 18 Mar 2003 10:55:13 +0100
| Newsgroups | gmane.comp.jakarta.avalon.phoenix.devel |
|---|---|
| Message-ID | <[email protected]> |
Nicola Ken Barozzi wrote:
> What is so eeevil about it? What makes javadoc tags as attributes much
> better?
Nono, I think that using javadoc tags for metadata just as hacky. It's
horrible.
Java got polluted the day when the compiler looked into javadoc tags to
find out if a method is deprecated.
SoC is not something the java architects were to clean about after 1.0
was nailed down.
This is why, at least, I would be proposing to *namespace* comments, so
that we have a easily parserable way to separate concern areas in the
syntax.
so
// --> single line comment
/* */ --> multiple line comment
would be general comments
/** */ --> javadoc comment
[there is no single-line javadoc comment]
/*+ */ --> metadata
[there is no single-line metadata comment]
//[...] --> single-line AOP
/*[...] */ --> multiple-line AOP
where [...] is the aspect associated with it
- o -
the above would, for example, solve the current cocoon issues with
making the flow and the instrumentation modular.
and could well be implemented with a code prepreprocessor.
but there is something in the back of my mind that keeps setting my
elegance alarms off and I don't know what this is.
One thing if for sure: the java syntax was not designed to be aspect
orientable. so every solution will be partially hacky. We have to accept
his until the new best-of-breed language comes around.
Stefano.