Updates: OpenJDK Forum: Contributing into OpenJDK 6

[Dalibor Topic]

Mark Reinhold wrote:
>> Date: Mon, 02 Mar 2009 18:34:08 +0100
>> From: dalibor.topic at sun.com
> 
>> Andrew Haley wrote:
>>> Hi there,
>>>
>>> is a recording of the call available?
>> Not yet - it should become available later today.
> 
> Here you go:
> 
>     http://mediacast.sun.com/users/mreinholds/media/openjdk-forum-2009-02-27.ogg
> 
> - Mark

And here are the rough meeting notes:

OpenJDK Forum - Contributing into OpenJDK 6

Date: 2009-02-27

Participants: 22

Speakers (in order of first appearance):

Dalibor Topic
Mark Reinhold
Joe Darcy
Andrew Haley
Dmitri Trembovetski
Lillian Angel
Martin Buchholz
Phil Rice
Volker Simonis

Discussed items (in slightly adapted order):

* OpenJDK 6 & writable repositories

We have the OpenJDK 6 writable repositories. Marting Buchholz made 
some pushes, and we'd like to see more people do it. Potential 
patches to  include in OpenJDK 6 would be IcedTea pacthes, security 
fixes, backports of fixes from OpenJDK 7.

Ports like Zero or Shark are larger then simple backports, so they 
should go upstream. In order to make that happen, there is a desire 
to understand processes, and get up to speed as quickly as possible.

* How do we go about merging in larger changes like Zero?

 - A large change like that should go into 7/6 simultaneously, or into
   7 first.
 - Bugzilla entry for Zero
 - Create webrev
 - upload it on cr.openjdk.java.net server
 - Send it to HotSpot team for review
 - have someone mentor the external developers on the first merging steps
 - have them spread around different areas of expertise

* Bugzilla

We have a bugzilla up now, at the moment its purpose is to collect 
patches from developers without push rights. Simple process for
contributions, and a draft of sponsorship process - mainly of interest
to Sun engineers right now. Dalibor and Mark are poking Sun engineers
to look at them.

* What's the procedure for committers for a simple patch from IcedTea
into OpenJDK 6? 

 - generate webrev of the change for large change, diff is fine for small ones
 - post the patch on jdk6-dev and ask for review on jdk6-dev alias
 - approval from technology expert in the area (Joe and potentially others)
 - backport from OpenJDK 7 will usualy be appropriate
 - push the change into OpenJDK 6

* What's the process to get push rights into OpenJDK 6 ?

- start with a few patches with some assistance
- figure out the process with the mentors
- get push rights

* What's the merging activity planned for OpenJDK 6?

 - not much - some of the JDK 7 bug fixes planned for backporting and merge by Joe.
 - Martin expects 10 changes from him/Google he'd like to see in OpenJDK 6

* Can the backporting activity be delegated on more shoulders?

- Joe would be happy for other people to do the backports and pushes
- it's more of an issue to identify desirable fixes
- once a fix is in 7 it's not very hard to backport into 6

The client team started a big push to forward port changes from 6uN into 7.

* Should we use bugzilla to track merging activity?

 - gcc uses bugzilla for merging larger patches

* Should we have small changes go through the list, and large changes through
  bugzilla?

 - Joe, Lillian are OK with that approach.

* what areas to work on OpenJDK7 without duplicating effort ?

- a good first batch of stuff is the small patches from IcedTea
- and then have the conversation again
- probably about Zero & the plugin

* How about the IcedTea plugin?

 - it can likely be contributed under the SCA
 - it may make more sense for OpenJDK 6, then for OpenJDK 7, where Ken's plugin will appear
 - it can be replaced later if the new plugin is better
 - Sun isn't necessarily going to build & test it, but that's OK 

* what else do people want to work on?

 - porting
 - client-side speedups and size improvements
 - C++ interpreter (and tiered compilation)

* Cultural differences in bringing out the review process into the open

 - A struggle we expect to continue to have for a while, but Sun 
   engineers are coming to understand that it's a good thing

* How to go about encouraging reviews to happen and to happen in the open?

 - hard to get past the known sub-group trust relationship regardless of sub-project
 - should we use reviewboard - we need to review it and put up an instance
   for experimental use by early summer
 - webrev robot from some sun teams, but reinforces sub-group trust 
   relationships, and not hardened for life on the internet.
 - open source review tools usage not common in practice, but experience shows 
   reviews are useful

* Relationship and patch flow between OpenJDK 7 and various internal releases?

 - The vast majority on hg.openjdk.java.net
 - Small bit of left over encumbrancies for 7 internal at Sun
 - No internal 7 tree with a different code tree
 - The release engineering pulls every night from the hg to get the open code,
   and from the closed bits to turn them both into 7 builds
 - pushes to OpenJDK 7 will appear in the next 7 builds
 - a few build variables decide for some encumbered bits of code which 
   one to use

* Can SCSL licensees push directly without signing the SCA?

No. There is a process to take SCSL-covered code and turn it
into something that can be pushed into OpenJDK. 

The forum was successful, so we'll have another one in a few weeks time.

cheers,
dalibor topic

-- 
*******************************************************************
Dalibor Topic                   Tel: (+49 40) 23 646 738
Java F/OSS Ambassador           AIM: robiladonaim
Sun Microsystems GmbH           Mobile: (+49 177) 2664 192
Nagelsweg 55                    http://openjdk.java.net
D-20097 Hamburg                 mailto:Dalibor.Topic at sun.com
Sitz der Gesellschaft: Sonnenallee 1, D-85551 Kirchheim-Heimstetten
Amtsgericht München: HRB 161028
Geschäftsführer: Thomas Schröder, Wolfgang Engels, Dr. Roland Bömer
Vorsitzender des Aufsichtsrates: Martin Häring