[foreign-abi] On invokers

[Samuel Audet]

On 9/25/19 8:32 PM, Maurizio Cimadamore wrote:
> I'd like to point out the irony of your comment: you are complaining 
> about lack of transparency on a thread which is discussing future 
> directions for SystemABI support (where the currently proposed patch 
> only works for Windows, only for downcalls). Speaking about open-ness, I 
> think this is pretty much the best you can get?

Which is great for these parts, like I said, but the same isn't 
happening for other parts, such as anything related C++. Maybe it's 
because you do not have time? I don't know! You're not being transparent.

> I will disregard your comment on "you should do X instead of Y" - which, 
> as been pointed out previously, is not helpful, and I will also 
> disregard your "framework-ish APIs" remark, which is deliberately 
> inflammatory. I think we can do better than that?

I didn't mean it as inflammatory, just that it doesn't fit in the 
API/implementation separation that you're striving to achieve. I could 
elaborate, there are many technical issues with your approach, but since 
you haven't heard me out in the past, I just don't have the feeling that 
you'll hear me out this time around.

> As for vectors, I believe the discussion John is referring to 
> happened/is happening on the CSR front:
> 
> https://bugs.openjdk.java.net/browse/JDK-8223348
> 
> (again, discussion publicly available).

Thanks for the link! That's news to me. As usual, for anything related 
to vector APIs, that's looking great.

> And yes, there are times where, as normal people do, we need to stop and 
> think about what we wanna do before starting a public debate on it. This 
> is about the evolution of the Java platform, not about conducting 
> opinion polls (although sometimes it helps to have dedicated surveys on 
> specific topics, as we have done in the past [1]) - and it is a job that 
> we take very seriously.

I have no doubt of your seriousness, but I do hope you realize one day 
that accepting ideas from outside may be more effective in helping out 
with thorny issues.

Samuel

> 
> Maurizio
> 
> [1] - 
> https://mail.openjdk.java.net/pipermail/amber-dev/2018-February/002669.html
> 
> 
> On 25/09/2019 11:35, Samuel Audet wrote:
>> Hi, John,
>>
>> I must say, I really like the work you guys are doing on the vector 
>> and ABI fronts, it's really great. I sincerely wish Panama could focus 
>> more on that instead of also working on higher-level framework-ish APIs.
>>
>> I have one question though. Where are those "vigourous conversation" 
>> happening? I don't have any quarrel with the work being discussed on 
>> this thread, but it does give me the impression that a lot of the work 
>> related to Panama is done in secret and that we're not getting the 
>> full picture, about other things that I worry about like C++. If 
>> that's the case, that's fine, OpenJDK doesn't need to be "open", but 
>> it would help the community if you could be honest about it.
>>
>> Samuel
>>
>> On 9/25/19 9:25 AM, John Rose wrote:
>>> (*Mature*  non-generalization appears when you have a road map with an endpoint, so you can say things like,
>>> “You might ask my API to do extra task X, but you should really use a different API for that.”  Folks vary by
>>> temperament which kind of premature-ness they are more likely to fall into.  Y'all know I’m an (a), so of course
>>> I’m always on the lookout for signs of (b), whether they exist or not.  The Vector API is currently having a
>>> vigourous conversation about such choices.)