Bytecode generation, Source code mappings, JCov, Future (Patch)

[Jonathan Gibbons]

Alex,

As a prototype, you could also create a project on dev.java.net and  
publish your work there, for people to try out and evaluate. We could  
link to your project from the OpenJDK compiler pages as well, if you  
wanted.

If you start getting more traction with the idea in OpenJDK, it may be  
appropriate to create an OpenJDK project for the work. OpenJDK
projects have to be sponsored by OpenJDK groups, and right now, my  
sense is that this work is a better fit for the Serviceability group,
if they're interested. See http://openjdk.java.net/groups/serviceability/ 
  for info on what they cover.    If they were interested, the  
compiler team would obviously still be involved to deliver any  
necessary information from javac.

-- Jon

On Apr 23, 2008, at 8:49 AM, Alex Rau wrote:

>
> On 23.04.2008, at 17:17, Jonathan Gibbons wrote:
>
>> Alex,
>>
>>>>> However it should be sufficient to discuss the implementation.  
>>>>> As soon as the patch is acceptable I'll submit the SCA.
>>
>>
>> I'm sorry, it doesn't work that way. For legal reasons, we cannot  
>> look at your patch until you have signed the SCA.
>>
>
> I've submitted the SCA yesterday via Fax (sorry, forgot to mention).
>
>> Separately, although the work sounds like it has potential, I don't  
>> think it is appropriate to consider it for submission at this time.  
>> You are proposing adding a new output file format for the compiler,  
>> which is a step that should not be undertaken lightly. I'm not  
>> saying it needs anything as big as a JSR, but the file format  
>> should at least go through some sort of design review by the likely  
>> consumers of such a file. If nothing else, the Serviceability team  
>> should be involved, so that other JDK tools which might want to use  
>> the file are considered.
>>
>> I've started a thread on serviceability-dev for anyone interested.
>>
>> -- Jon
>>
>>
>
> I'm aware of that and I think you are totally right. My intention  
> with the patch was more kind of "proof-of-concept" and idea  
> brainstorming how the feature could be implemented - especially as I  
> agree with you that a formal specification/discussion basis like a  
> dedicated JSR would probably be overkill. This (the patch) seemed to  
> me a "more" appropriate way of talking (or at least getting started  
> talking) about the "same thing". Please just consider it as a  
> prototype which has nothing to do with the final implementation nor  
> it was intended to be targeted for  "submission and final  
> integration" ( although I must admit that I am not familiar yet with  
> the official submission process).
>
> Best Regards,
>
> Alex
>
> PS: Thanks for getting more people involved
>
>