No 0.16.1, but quick 0.17?

[Mark Wielaard]

Hi,

I don't think we need a branch or backporting fixlets to 0.16 for small
bug fixes. We are in rapid development mode for various parts of our
library. If people want we can make a new 0.17 release at the end of
this week (next weekend) that incorporates all new work done since 0.16.
It is a good idea in general that people have the feeling we are always
"ready to release". And I would like to see clear progress and not
various branches and forks at this time. And creating new versions costs
real (non-trivial) maintenance and testing time. Time that I rather
spend on pushing out real new versions.

So if people feel we need a new release soon then here are some extra
goals for 0.17:
- Get XML DOM fixes in.
  Done. Chris, I assume you want to see all other fixes also deployed
  soon.
- Update integration manual.
  Andrew Hughes posted a RFC:
  http://lists.gnu.org/archive/html/classpath/2005-06/msg00104.html
- swing busy paint/validate loops cleanups
  The removal of paint() and revalidate() has improved the feel of free
  swing a lot and although there are some small redrawing issues I think
  this is a big improvement on usability over 0.16. Roman, do you feel
  this is an improvement? What other fixes do you think are essential to
  fixup any remaining drawing glitches before 0.17?
- gcj 3.4 miscompilation (not found yet)
  This was reported on irc. The GNU Classpath Examples Swing Demo
  doesn't receive any button events. If at all possible I like to find a
  workaround for this issue since I want to switch to gcj 4.0 soon. It
  would be nice to have the last release that supported gcj < 4 actually
  works for our demos.
- has someone looked at this error? I see it a lot when trying out
  various swing examples, but don't have a sure way to replicate yet:
  Exception during event dispatch:
  java.lang.ClassCastException: javax/swing/JMenuItem
    at java.awt.DefaultKeyboardFocusManager.dispatchEvent
       (DefaultKeyboardFocusManager.java:197)
    at java.awt.EventDispatchThread.run (EventDispatchThread.java:74)
- when zip not found classes are not compiled/installated?
  Reported by Stephane for cygwin.
- There were some hangs reported with eclipse. I have seen this happen
  myself now, although I think this is not new. It looks like our io/net
  can get interrupted somehow and doesn't recover correctly.
  (look at libgcj CNI code to see if they deal with this better then our
  JNI code).
- In general get all the new cool stuff already done (and being done
  this week) ready to rock! Corba updates, ObjectStreamClass fixes,
  build/configure fixes, FSF address change, Swing/AWT documentation
  updates, jtable fixes, etc, all already done since 0.16 was released.

What do people think?
Please work under the assumption that we can release a new version at
any time. So think whether you would be embarrassed by a release that
ships with the code that you are about the check in. If so, don't :)

Cheers,

Mark

-- 
Escape the Java Trap with GNU Classpath!
http://www.gnu.org/philosophy/java-trap.html

Join the community at http://planet.classpath.org/

signature.asc
Description: This is a digitally signed message part