Re: [PROPOSAL] lifecycle release

On Thu, 20 Mar 2003 15:14, Niclas Hedhman wrote:
> > Metadata makes the components easier to use, deploy and maintain. You
> > could store it outside the components in config files but then you ha=
ve
> > to make sure multiple artefacts keep in sync. Having the config files
> > generated from the source (or baked in) makes maintainence simple. Be=
st
> > place to document code is with the code and metadata is just aform of
> > documentation.
>
> We live in an XML world (or at least I do), and having a "unified" XML =
Meta
> description document, that can be transformed (for compatibility reason=
s?)
> to each containers requirement, sounds a lot easier to me...

Doesn't to me.=20

The trend of technology is to add them in. Witness C# and its=20
incorporation of ADD, witness the success of xdoclet, ejbdoclet etc (Do y=
ou=20
know of anyone who still hand codes EJBs?). Most development technology h=
as=20
been moving towards self documenting code.

> Especially considering the @tags already in some of my source, from UML=
 and
> JDO (Kodo). It is outright ugly, and after specified (hardly every up f=
or
> change), just occupies vasts amount of precious screen space. Easily ge=
t
> half the screen of @tag lines.

so - Its a crap implementation then ;)=20

Have you looked at ADD (either XDoclet or C#)?

> Also, if you are requiring me to have dedicated Ant tasks (not everybod=
y
> builds with Ant (strangely enough)) to be used when I develop my Avalon
> blocks/components, then you ARE raising the bar significantly.

whether it is ant tasks, commandline tools or maven plugins most people u=
se=20
tools of one sort or another so I don't see that as a major barrier of en=
try.=20
In JDK1.5 it will be part of javac so then there will be 0 changes.

--=20
Cheers,

Peter Donald
*----------------------------------------------------*
|    "the mother of idiots is always pregnant."      |
*----------------------------------------------------*