Skip to main content
Jump to: navigation, search

Eclipse/PMC

Documents

Some documents written and/or used by the PMC:

Meeting Schedule

The Eclipse Project PMC has a weekly phone meeting every Wednesday at 10.30am EST.

Meeting Minutes

September 30, 2015 - McQ, John, Alex, Dani

  • Dani: will send a note to PMC list asking to approve new Debug leadership (Sarika)
  • Dani: we should finalize our API removal discussion from last week
    • agreed that APIs marked for removal have to be annotated with @noreference
    • agreed that components should be allowed to remove API but they have to provide good reasons
    • agreed that we won't allow to delete APIs simply because they are deprecated
    • agreed that the PMC will decide case by case i.e. there will be no general rule
    • regarding version numbering we decided to also decide this case by case
    • Dani to update the removal document and have it reviewed by the PMC

September 23, 2015 - Dani, John, Alex, Martin

  • Dani: JDT Core - Co-lead going to step up
  • Dani: API Removal Discussion
    • Q1: When do we actually delete API? What's the benefit compared to the pain that we cause ?
      • Example of methods that don't do anything any more or do wrong things -- those should be removed
      • Example TableTreeViewer : Continue having the API doesn't hurt, there's no significant benefit removing it
        • Alex: TableTree was completely broken on GTK for 2-3 years ... keeping such components that don't work properly lowers the quality
        • Dani: Is there actual proof of bugs ? Or could it be working fine on Windows RCP ? If it's deprecated, people use it at own risk; do we really need to break them, if it provides value to some people on some Platforms ?
        • John: In TableTreeViewer case, EMF had some generic code (was unclear if the path was ever taken) and CDT could update easily
      • Summary: scheduling for removal is OK with good arguments. Give Adopters a chance to respond before removal takes place.
    • Q2: How to deal with the versions?
      • Dani: Updating the major causes major pain on everyone (adoption work), so this should be avoided
        • Actively developed plugins will notice source breakage when recompiling anyways -- no need to update the major for them.
        • For dormant plugins (not recompiled), everyone will break when updating the major although only few may be affected - is it worth notifying those small percentage that might break ?
        • Plugins who don't care recompiling may have to live with ClassNotFoundException
        • Tooling exists: API Use Scan Tools can discover incorrect API references that are not announced by the versions
      • Summary: Handle the Major with care -- in most cases, the cost of updating the major is not justified by the benefit.
    • John: Announcement When thinking about removing something, we should announce that far and wide and ask for feedback
      • Martin: But which channel is as effective as actually removing it ? There's always who don't actually listen...
      • John: Still, giving a possibility to listen is important. Agree that mentioning in the release docs is not enough.
      • Dani: When making a release, also send message with a link to the removals page (for all removals that are planned)
    • John: Mechanisms for maintaining binary compatibility while only breaking source compatibility (but it's a lot of work!)
      • Dani: Agree, in this case better just leave it in there
    • Alex: What to do next time, can we remove more stuff ?
      • Martin: Should be at the discretion of the committers. They do the work. If they see the need for removal, they should be allowed to do so (as long as they play by the rules, like early announcement). Need to define what the rules are.
    • John: There was an interesting discussion on cross-project, asking for well-known points in time where major breakage can occur
      • Eg release but without all the deprecated at certain well-known point in time eg every 3-5 years
      • AI continue that discussion on the Architecture Council
    • Summary: Essentially do what we did, plus more communication upfront, allow people to respond before deletion happens (to avoid churn)
      • Committers still need to be able to delete stuff when they find it necessary.
      • Updating the major (or not) to be decided case by case, but in many cases "breaking everyone" is not justified against "notifying few dormant plugins".
  • Alex: Bumping the minimum GTK version again (may cause issues on Platforms like AIX -- to be discussed when it's time)



September 16, 2015 - John, Martin

  • John: API Removal Discussion
    • No urgency now -- changes have been reverted for now, and scheduled for 2017
    • Updating the major of a bundle knowingly breaks everyone/most adopters
      • In the past, breaking changes have often been small enough to work without increasing the major
      • One can argue that removing TableTreeViewer is big enough to warrant updating the major
    • Versioning packages has not been done in the past due to the huge upcoming maintenance effort when starting to do so
    • "Release Version" is decoupled from "bundle versions" already (and may move to date-based versions eg "2016.1" with rolling updates moving forward
    • --> will have more discussion next week

September 9, 2015 - Dani, Alex, Martin, John, McQ

  • John: Planning Council Updates
    • 4 planned releases (March, June, September, December -- essentially end of each quarter) with flexible contents
    • Mid December rather than end to avoid churn, so this one is a little shorter
    • Only June is "major" - allowing to drop off, or breaking changes; others are "minor"
    • McQ want to reduce the number of simultaneous streams -- if "master" is more stable more often that's OK, but avoid too many "live" streams
  • Software is getting more important - would be good to better support multicore
  • John: IntelliJ change in licensing / sales model
    • Many eclipse-positive comments on the announcement blog
    • Possibility putting Money on Eclipse Development may become interesting for companies in this context

September 2, 2015 - Dani, Alex, Martin, John

  • Dani: EclipseDay India on Saturday, 200 attendees wanted to join, hat to cut to 150
    • Keynote by Mike Milinkovich - large Community
  • Dani: Policy for and Mars.2
    • Do we want to stick to the "Service" model or allow feature updates ?
    • Mars.1 winding down -- sticking to "Critical Fixes Only" for that
    • Too much in the maintenance stream causes risk of defocus ... are there relevant features that are worth the extra effort ?
    • Dani: Suggests to require PMC Approval for adding a feature in - example candidate: Improvements for HiDPI
      • Also: What about version number (2nd digit version update), IP disclosures, Translations ... ?
      • Dani would suggest sticking to 3rd digit update only in the marketing release number; but a Release Review would be needed

August 26, 2015 -

  • Dani/Alex/Martin can't join (traveling)

August 19, 2015 - Alex, Dani

  • nothing to discuss

August 12, 2015 - John, Dani

  • John asked whether we run on Windows 10
    • Dani: yes, the team already tested on it a few weeks ago. Runs smoothly one bug so far. Browser widget works despite new browser (Edge)
    • Martin (added after the meeting): A CDT update is needed to keep the Terminal from hanging (see bug 474327, will release with Mars.1). Got some duplicates already. Workaround is switch the Win10 Console to "Legacy Mode".
  • Dani would like to get plan feedback by Friday EOD

August 5, 2015 - McQ, John, Dani, Alex, Martin

  • PC meeting later today (planning calendar, calling SR1/2 "Update 1/2" instead
    • adding another release before Christmas might be a next step - even if Platform contributes identical bits
  • Dani: Eclipse/Mars Retrospective
    • Move more components to Tycho build? (Would still need Ant to test against final build/bits)
    • Contribution Review Dates: joined by some components but not all
    • Error Reporter: Interesting to look at top ten but the sheer number is too big
      • John - based on Orion experience with similar error reporting :
        • Looking at changes in reported issues is more interesting than looking at reports themselves
        • Reports help getting contributions (But, Dani finds that "just adding a null check" is often not what's desired for Java .. though helpful for Javascript)
  • Dani: Foundation IP team doesn't require updating copyright notices per contribution any more (since that information is in git anyways)
    • The Project has to agree
    • Some contributors like to have their name in the source -- that's OK, no requirement to remove author information, but no requirement to add either
    • Won't remove existing lists (they never claimed to be complete, since there always was the "...and others" copyright notice
    • Dani to sent request for voting
  • Dani: Switching to Jetty 9.3.x (which requires JRE 8)
    • JRE 8 from Oracle (and also from IBM) exists for all Reference Platforms
    • Except Solaris, because we only support Solaris 32-bit and the JRE only exists as 64-bit
    • But the Plan for Neon is to have 64-bit Solaris support
    • New Language features in Java 8 are adopted, contributors would like to start using Java 8
    • McQ: In the past, staying on older Java was desired to enable more widespread use ... today, this argument does not seem valid any more, in fact likely more contributions / community is enabled by moving to Java 8
    • Dani: Only concern is some "non reference" Platforms like HP-UX might not have JREs initially; but that's OK as long as the reference platforms are good
    • AGREEMENT to move to JRE 8 and allow projects to use Java 8 in their code.
  • Dani: Looking for a contributor for SWT improvements for GTK3
  • Martin: libwebkitgtk-3 on Ubuntu 14 forcing GTK 2 not working ?
    • Alex: Using libwebkitgtk-4 which is much more stable, but not implementing the full SWT API
    • Most distros don't ship libwebkit for gtk-2 any more since it's not supported upstream any more and has many security issues
  • Alex: Build SWT at the Foundation
    • Work with the Foundation going well, expect to have RHEL machines deployed at the foundation next month



July 15, 22 and 29, 2015 -- no meeting


July 8, 2015 - McQ, Dani, John, Martin, Alex

  • John: Cross-Language Tooling Discussion on the eclipse.org-architecture-council and ide-dev mailing lists
  • decided to cancel the upcoming July meetings

July 1, 2015 -- no meeting


June 24, 2015 - McQ, Dani, Martin, Alex

  • Dani: Java 9 - <a href="http://marketplace.eclipse.org/content/eclipse-java%E2%84%A2-9-support-beta-mars">EAR Feature Patch</a> on the Marketplace NOW
    • No JARs any more - JRE is doing things internally using "jimage" format; updated search etc to create projects and work against them
    • If the Jimage filesystem provider isn't backported, one has to run the IDE on Java9 in order to code Java9
    • Modules are just a list of packages (and can refer to other modules) - no real JSR describing the plan yet - seems like just a replacement of "Profiles" (and JARs)

June 17, 2015 - Dani, John, Martin, Alex


June 10, 2015 - Dani, Alex, Martin

  • Dani: 4.5RC4 looking good, no more fixes planned
  • Dani: Working on Java 9 feature patch
  • Alex: PC discussing a change in the release train
    • Current common ground seems to be a request for more release points, and projects could decide whether they do features or maintenance
    • From Platform point of view, stability is key. Some key contributors not interested doing
    • Martin: How to also cater to contributors who want their contributions released soon ?
      • Martin Idea: With Tycho, building the Platform is easier so ask contributors build themselves
      • Or, open up a new "experimental" stream ?
      • Dani Idea: Market milestone builds differently, as "fully consumable" would serve the same purpose
  • Martin: Tested eclipse-installer (Oomph), looking really really good now

June 3, 2015 - Alex, Dani, Martin, McQ, John

  • Dani: Mars Endgame
  • Dani: Please vote for release review

May 20, 2015 - Dani, Alex, Martin, McQ

  • Dani: Security Update - Platform work done, Orbit updated, reached out to Wayne and other affected projects.
  • Dani: RC2 Build - too many bugs assigned, Lars on Vacation, Dani will fill in
  • McQ: Too Many Platforms Built? - Who's really hurt by "too many builds" ?
    • Will meet with Mike & Foundation tomorrow, Alex is also interested (Dani to check).

May 13, 2015 - McQ, John, Alex, Martin

  • John, Dani - Mars Endgame looking good
  • Alex - bug 465874 Lucene 5 looking good, almost done - Ready to commit as soon as CQs are in and Mars+1 is open
  • JDT for Java 9 - will need a wider discussion with EMO on make it easier to publish the work, e.g. in normal builds

May 6, 2015 - McQ, Dani, Martin, John

  • Dani: Java 9 timing - slight delay
  • Dani: RC1 preps
    • 2-day test pass went fine - 2 severe issues found, will be addressed
    • Request to watch PMC mailing list for API exceptions and defect approvals
  • John: PC Discussion on Release Cycles
    • Multiple releases per year PLUS maintenance streams seems like overkill
    • Consider an approach like Orion that just moves constantly forward
    • Especially for the Platform, being rock solid is most important. Still to attract new contributors we need to allow more frequent "feature updates".
      • A model where both "stable/maintenance" _and_ "features" are contributed to the train might be too much work/overhead.
      • Consider a model like Ubuntu, ... with some release numbers being "stable/LTS base" and others being "in-between feature releases" ?
      • Consider a model like LTS for maintenance fixes / aside mainstream just moving forward ?


April 15, 2015 - Dani, John, McQ, Alex, Martin

  • Dani: Java 1.7 Changes
    • Some bundles have been moved to a 1.7 BREE by new committers, even after API freeze
    • Rule has always been "we move up when there's a reason to move up". We won't move up without reason.
      • Dani: Moving the BREE may even cause API changes, so should only be done when incrementing the minor version (5% risk)
      • Alex: Such updates allow staying current and not get to "rewrite is needed" state (thus needed) but has to happen before M6 (API freeze)
    • Alex suggest not accepting additional changes, but not reverting either (to avoid churn)
  • Dani: Batik 1.6 update

April 8, 2015 - Dani, Martin, Alex, John, McQ

  • Dani: Batik - Platform is good, Train may need to update, perhaps updating one bundle only would suffice. John will follow up.
  • Alex: SWT for GTK 3 News
    • GTK port finally decoupled from X11 - it renders on Wayland now, can switch the renderer to a pure HTML one
    • This opens up opportunities (but depends on hosts that have GTK).



April 1, 2015 - McQ, Alex, Martin, Dani (Regrets: John travelling)

  • Alex: GTK 3.16 seeing issues again - fixed some crashes, but scrolling is still entirely broken
    • SWT uses a number of things that GTK declares as "implementation detail"
  • PMC approval on piggyback CQ's (AC question forwarded by John)

March 18, 2015 - Dani, Alex, Martin, McQ, John

  • John: EclipseCon - Bigger this year due to LocationTech (750 attendees)
    • Mark Reinhold keynote and "after-session" on Java 9
    • Much interest in Orion JS tooling / editor, also on desktop
    • Public face of Eclipse Platform at the conference was much more diverse than in the past (Lars Vogel, Max Anderson, Google ...)
  • Dani: bug 458730 Mars Plan Update
  • Dani: e4 project leadership approved by EMO
  • Dani: Szymon Brandys resigned as Platform/Resources co-lead. Need to +1 on the mailing list

March 11, 2015 - no meeting (EclipseCon)


March 4, 2015 - Dani, McQ, Martin, John

  • Dani: e4 leadership - Dani will volunteer to co-lead
  • Dani: BREEs - documentation about how to pick the EE
    • Recommending the "earliest generally supported JRE that provides the capabilities you need"
    • Would like an URL on the page pointing to the most recent plan (talking to Wayne)
  • John: greatfix contest
    • Dani: Working well - some very small contributions but some also very large (eg Customize Perspective fixes)
  • John: EclipseCon - numbers looking good; join Planning Council Breakfast as delegate for Dani

February 25, 2015 - Dani, Martin McQ

  • No topics

February 18, 2015 - Dani, Alex, Martin

  • Alex: Building Native Launchers
    • Current way of building is kinda unpredictable - even if getting some agreement on versions to use, results are kinda unpredictable
    • Pushing towards Hudson RHEL builders at least at the EF to get more transparency and automation - attempt to mimic the infrastructure at IBM
    • Looking at 3 primary architectures (at the EF) for Linux vs. secondary architectures (non-public builders potentially)
  • Dani: Great initiative, but other (non-EF) builders must not be broken
    • EF doesn't allow any commercial tools (but currently, e.g. Windows launchers are built with MSVS)
  • Alex is willing to spend time to get Linux builds running; but can't help with other architectures
    • Martin: great approach - for Windows, using a cross-compiler on Linux might be interesting (after Linux native works)
  • Alex: This is just phase one - getting rid of the binaries in git repos might be phase 2 (since the checked-in binaries easily cause inconsistencies between Java and Native side)
    • Martin: Checked-in binaries help consumers and contributors who just want to make a Java change
    • Dani: Checked-in binaries are also used for comparing build results for expected vs accidental changes
  • Alex: bug 459399 - Policy for recommended minimum execution environments for bundles
    • Dani: It works today
      • To run Eclipse, Java 8 or Java 9 can be used (minimum BREE has no impact)
      • To modify the source, a new JRE can be used but then the Execution Environment Descriptions need to be installed
    • Policy as discussed in the past: Each project can increase the BREE if there is a real need (such as generifying) and no upstream clients are broken
      • But don't change the BREE without justification -- changing the BREE always has some effect, such as new warnings that would need to be addressed
      • Suggested BREE for new bundles has already been changed by Lars
  • Alex: Even for bundles in "maintenance mode", old BREE causes issues for people who build from source (who have to change compilers etc)
  • No conclusion so far (Alex and Dani disagree)
  • Dani: e4 leadership
    • Mature bits being moved to Eclipse - e4 remaining as an incubator to keep alive for experiments with low entry barrier

February 11, 2015 - Dani, Alex

  • no topics

February 4, 2015 - Dani, Alex, Martin


January 28, 2015 - Dani, Alex, John

  • Dani: Switch Mac OS X 10.9 with 10.10 in Mars target environments
    • No objections
  • Alex: Looking for any Eclipse related activity @Fosdem

January 21, 2015 - Dani, Alex, Martin, McQ

  • Alex: Process for allowing non-committers extended bugzilla privileges (for bug triage)?
    • Dani: Yes a process exists. Send bugzilla username to Dani.
  • Alex: New resource for helping with SWT
  • Dani: Platform/UI co-lead
  • Dani: Solaris: Java 8 will only support 64 bits on both Intel and SPARC --> IBM SWT Team considering to invest in getting patches in for 64-bit Solaris

January 14, 2015 - Dani, Alex, Martin, McQ, John

  • Dani: Update on Platform/UI Leadership: Daniel Rolka left IBM and for now has no time to contribute. He stepped down as co-lead and nominated Lars Vogel
  • Dani: Solaris x86 64-bit support - patches exist, but no machine available. No luck to get one from Oracle or via Eclipse Foundation. We will not support Solaris x86 64-bit unless someone makes a machine available

January 7, 2015 - Dani, Alex, Martin, McQ, John

  • Dani: Platform/UI Leadership
  • John: Git security issue - pick up a patch for Jgit in the packages before SR2? - Mostly an EPP

Archive

Back to the top