[jsr294-modularity-eg] Runtime model
Andreas.Sterbenz at Sun.COM
Thu Jun 7 03:44:17 EDT 2007
Bryan Atsatt wrote:
>> Of course it is a collaborative process.
> If so, is it unreasonable of me to expect a response to the model I
I responded to the parts that I saw issue with. For the rest I did not
have major comments, so it did not seem necessary to respond one by one.
Silence does not imply disapproval. The best way to examine those areas is
to contrast them with the proposal Alex and I will send out, probably
early next week.
>> All I tried to say is that Alex and I have already been working on a
>> complete proposal for the reflective API for some time. It will soon be
>> ready and when it is, we will send it out. If you want to send out a full
>> proposal of your own, you are welcome to do so.
> So... we can't discuss models in an informal way, only via full proposals?
It is easier for me (and I assume the rest of the EG) to evaluate a
proposal if its ideas are fully formulated and no major areas are left
blank. I don't think that is an unreasonable bar for proposals to the EG
to meet, whoever they come from. That includes myself and you are welcome
to hold me to it.
Also, Sun's responsibilities as spec lead include producing a
specification, a reference implementation, and a TCK. If we don't
understand a proposal to sufficiently to produce them, we cannot accept
it. That means if you make a proposal all of your own, you must commit to
working with us to get it into a shape from which we can take it the rest
of the way. Rights and responsibilities.
> I understand. But there will be a long transition period in which legacy
> code can be wrapped and exposed as modules. Glyn's use-case can easily
> be extended to include this one.
I assume by modules you mean JSR 277 modules. Given that JSR 277 and 294
are both targeted for JDK 7, you cannot run modules on earlier JDK
releases irrespective of JSR 294.
One could try use a JSR 277 JAM file as a plain JAR file, but you'd lose
all the benefits of the 277 runtime infrastructure. Is that what you are
suggesting? Do you think that would be a popular option?
More information about the jsr294-modularity-eg