[OpenJDK 2D-Dev] RFR: 8184135: Remove obsolete dga code and binaries from Solaris SPARC build.

[Prahalad Kumar Narayanan]

Hello Phil

Thank you for the descriptive email explaining the changes.

As I understand, 
    . The proposed changes remove the usage of Solaris specific "Direct Graphics Access" extension in Java2d code.
    . I imported the code onto Ubuntu box (hope this is fine as both Solaris & Linux share same codebase) and the code builds fine.

The changes look fine to me.

Thank you
Have a good day

Prahalad N.

-----Original Message-----
From: Phil Race 
Sent: Wednesday, August 16, 2017 2:09 AM
To: 2d-dev
Subject: [OpenJDK 2D-Dev] RFR: 8184135: Remove obsolete dga code and binaries from Solaris SPARC build.


Webrev : http://cr.openjdk.java.net/~prr/8184135/
Long ago JDK worked best on certain Solaris SPARC framebuffers by using Solaris DGA.
This relied on a native library (libsunwdga) compiled from JDK sources together with some imported framebuffer-specific proprietary binary libraries pre-compiled for SPARC v8 & v9.

All of the "isDgaAvailable" and DGA locking code in the open sources are there only to support that.

However Oracle JDK9 has a silly state of affairs where it copies the framebuffer-specific proprietary binary libraries into the build but no longer builds the libsunwdga needed to work with them because JDK 9 does not support Solaris 10 .. the latest Solaris version which still supports these framebuffers on what is now mostly obsolete and unsupported hardware anyway.

So it makes sense to stop copying those libraries (not an openjdk
problem) but also to remove the
dead code that no longer is built. This has some ripple effect into X11SurfaceData where the locking code for DGA should then be removed as well as it demonstrably useless.

This code would only be interesting for the future if some very (very) similar looking support and even if the XFree86 DGA extension were interesting I don't think it is sufficiently similar to warrant keep the relatively few lines of around that might be relevant.

-phil.