[OpenJDK 2D-Dev] Openjdk java2d rasterizer JEP for pisces (marlin) enhancements ?

[Mario Torre]

2015-02-19 15:25 GMT+01:00 dalibor topic <dalibor.topic at oracle.com>:
>
>
> On 19.02.2015 13:50, Mario Torre wrote:
>
>> The code is part of an OpenJDK project, though, it's already the
>> existing Java rasterizer.
>
>
> It's not part of an OpenJDK Project - Project with an upper case 'P'. See
> http://openjdk.java.net/projects/ for details. Considering the motivation
> for this discussion, that would be a necessary precondition.
>
>> Of course I'm fine with giving it a formal home in the Graphics
>> Rasterizer project (where it naturally belongs), but let's also try to
>> be a bit more pragmatic, this code is in hold since more than one year
>> now, the JEP process can be started before giving a formal home, or at
>> least concurrently to that.
>
>
> While the JEP process is explicitly open even to aggressive,
> outside-the-box, and even completely wacky ideas, new feature ideas often
> require significant up-front research, experimentation, and socialization
> before they're ready to be proposed as enhancements to the JDK itself.
>
> That work is typically performed within an OpenJDK Project.
>
> That does not preclude writing any number of JEPs as a finger exercise, of
> course, but one should not expect that writing JEPs would in any form be a
> substitute for going through the significant up-front work necessary to
> research, experiment and socialize the desired changes within the OpenJDK
> Community.

While I agree with your general feelings, one has to start
*somewhere*. There have been 3rd parties doing extensive test on this
code already, and while those test need to be confirmed this is no
different than any other work done in the rest of the platform. Ah,
and yes, IcedTea will integrate it if it make sense, of course.

What we are discussing here is not a "aggressive, outside-the-box, and
even completely wacky ideas", is a set of patches that were partially
reviewed in very positive terms, but then put in limbo over the years.
Unfortunately, given the area I recognize there are not many
experienced developers who can comment on the code, but still,
ignoring that completely isn't fair either.

> For example, it would be interesting to see if, once the code has found a
> home in an OpenJDK Project, you, Mario, would take the time it takes to add
> it to IcedTea, and confirm that it does not cause issues with your
> regression and other tests on the various platforms you support. Among other
> things, that would allow you to form a more informed opinion about the code,
> potential integration paths, etc. and inform others in such a Project about
> your experiences.

I don't work directly on IcedTea, and afaik there is no IcedTea for 8
onward (afaik, but Andrew Hughes and Omair may tell us more on that).
Also, IcedTea is *not* a test repository for OpenJDK, is a stable
implementation of it, similar to what Oracle JDK is. That said, I'm
sure the IcedTea maintainers will do their part, especially if the
code turns out to be of the quality we are supposing.

Btw, a big chunk of this code, as I said, was sent out for review
already. If the reviewers thinks the quality doesn't justify the
efforts of integration into OpenJDK, then, please, just say so!

> Such work would have to take the time it takes either way. I believe that
> focusing on writing JEPs before such work is complete (or even begun) means
> setting one's priorities in a way that puts the cart before the horse. You
> can certainly chose to do that, but the cart might not get very far on its
> own. [0]

Ok, so let it not be a JEP. Let it not be a separate patch set for
review. What other options are there? I don't think discussing forever
whether we want a Project, a JEP or whatever means to put an horse in
front of the cart, it just means to park the cart behind a wall.

Cheers,
Mario

-- 
pgp key: http://subkeys.pgp.net/ PGP Key ID: 80F240CF
Fingerprint: BA39 9666 94EC 8B73 27FA  FC7C 4086 63E3 80F2 40CF

Java Champion - Blog: http://neugens.wordpress.com - Twitter: @neugens
Proud GNU Classpath developer: http://www.classpath.org/
OpenJDK: http://openjdk.java.net/projects/caciocavallo/

Please, support open standards:
http://endsoftpatents.org/