[jsr294-modularity-eg] Module directive types

[Bob Lee]

On Fri, May 8, 2009 at 9:00 AM, Peter Kriens <peter.kriens at aqute.biz> wrote:

> I can find myself quite well in this approach :-) I like the potential of
> type safety and constants a lot.
>

Thanks!


> Just some minor issues.
>
>    1. The ':' instead of '=' looks good, but seems to deviate rather
>    strongly from Java style? Though I must admit that it brought back pleasant
>    memories from my smalltalk days, but then the semicolon threw me off ...
>
> I can go either way. The intention here was to encourage a more fluent
style. For example:

  requires "foo" at: "5.0";

vs.

  requires "foo" version = "5.0";

We also use a colon in the map literal proposal, so this wouldn't look so
out of place.


>    1.
>    2. Would there be a possibility to allow syntax for a map? We have a
>    number of places where we have arbitrary attributes. Almost any language
>    supports a map type, never understood why Java did not have one. For example
>    [ a=b, c=d ]
>
> Absolutely! There's already a syntax for Map literals in Project Coin.


>
>    1.
>    2. I am a bit uncomfortable that we highly leverage annotations but add
>    a bit and deviate a bit. Over time, this will be a maintenance hell.
>
> Agreed. We must weigh the benefits of new features against the costs of
deviating. My suggestions were just meant as brainstorming.

>
>    1.
>    2. A lot of the ideas seem very interesting for annotations anyway, it
>    would be a shame if these features could not be used for other purposes. How
>    do we control this evolves together?
>
> Agreed. If we can add features both here and to annotations at the same
time, there's no deviation cost.

Bob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.oswego.edu/pipermail/jsr294-modularity-eg/attachments/20090508/9204ab5e/attachment.html>