Suggestions for Java Generics Semantics in Java Version 7.0 _ Resurrected

[Alex Buckley]

Hi David,

I know what you're getting at, but I don't think #3 implies a "Sun 
provides X, and everyone will like X" position. As a spec guy who spends 
more time in JCP Expert Groups than OpenJDK Project Groups, I am very 
well aware that the Java platform is not governed solely by Sun.

This is why, whenever anyone asks (demands) that so-and-so a feature 
should be added to the language and/or VM - the core of the platform - I 
wonder loudly if they have consulted other major stakeholders, experts, 
and enthusiasts. Perhaps the membership of the JCP SE/EE Executive 
Committee - Josh Bloch is not hard to reach. Perhaps our friend Dr 
Gafter. Perhaps the numerous individuals who gave very detailed feedback 
on the proposals sent to Project Coin.

I would be delighted if an OpenJDK project designed and implemented 
reification. As I said, it doesn't rank highly in Sun's own list of 
features, but it may rank highly in the lists of others.

But I would be even more delighted if that project's owners also 
explained their approach to reification - with pros and cons - to the 
global Java community along the way. I genuinely think people would be 
lukewarm about a project that, say, decided to ban raw types.

This outreach is one of the harder tasks in evolving the language. No 
doubt Sun itself, as spec lead for JSR 14, could have done a better job 
evangelizing migration compatibility back in the day.

Alex

David Herron wrote:
> Hey Alex,
> 
> This is a very well put argument about the situation.  I certainly don't 
> have the basis for commenting on the technological underpinnings of 
> anything you have said. 
> 
> The thing I want to point to is that item #3 follows a "Sun provides the 
> technology, you guys consume the technology" attitude.  In the spirit of 
> an open source community process perhaps #3 could be reworded to say 
> that Sun's priorities are clear from where Sun is placing it's R&D 
> dollars, and if a group were to form in the community to implement 
> reification ...etc... 
> 
> In other words, notwithstanding the prestige of the names you listed in 
> point #2, it may be possible with enough work for some set of people to 
> implement reification.  I would hope that if the community were to do 
> it, that it could be accepted into the OpenJDK.  Maybe some people in 
> the community have a different prioritization of needs than Sun has?
> 
> - David Herron
> 
> 
> On Wed, Jun 17, 2009 at 12:21 PM, Alex Buckley <Alex.Buckley at sun.com 
> <mailto:Alex.Buckley at sun.com>> wrote:
> 
>     1) Sun's bug database categorizes an entry as either a defect or a
>     request for enhancement. Bug 5098163 is, rightly, a request for
>     enhancement. It is not a defect that type erasure exists. I will say
>     that again. Type erasure is not a bug. It is a feature.
> 
>     2) The reason it is a feature is that many experts - including but
>     not limited to Philip Wadler, Martin Odersky, Gilad Bracha, Joshua
>     Bloch, and Neal Gafter - worked for many years on crafting a generic
>     type system that allowed people to migrate their non-generic clients
>     and libraries smoothly. Reified generics would have made that
>     migration more complex. If you don't know why, find out.
> 
>     3) While there were good reasons for avoiding it, there is no doubt
>     that reified generics make the Java language itself more regular. No
>     doubt at all. The question we face TODAY is, are reified generics
>     MORE valuable than other features that Sun can add to the Java
>     platform. Evidently not, since Sun is working on other features.
> 
>     Again, I don't want to prolong this conversation, but perhaps you
>     could say which JDK7 feature(s) you would drop to make room for
>     reified generics? You can choose from the list at
>     http://openjdk.java.net/projects/jdk7/features/
> 
>     Alex
> 
>