Minutes for Foundation IRC Meeting of September 12th, 2012

Meeting Log

Attendance

  • For the board:
    • Emmanuele
    • Joanie
    • Tobi
    • Shaun
    • Andreas

Agenda

  • Role and scope of the Release Team
    • Continuation of the discussion from the AGM at GUADEC 2012.
    • http://www.0d.be/2012/08/03/guadec-2012/

    • http://people.gnome.org/~ebassi/01-release-team.pdf - slides deck from the AGM

    • Clarification of the release rules and scope
    • The discussion at the AGM had to be interrupted for time constraints
    • The community considers the release-team as being in a good position to ensure the cooperation between maintainers in order to release GNOME
      • Some members of the community also would like to have an "arbiter" to resolve eventual conflicts
      • The arbiter could be the board, the release team, or a new team
    • Examples of issues faced by the release team
      • Deferring the python3 dependency
      • Acting as support for modules, by promoting them to the official moduleset, and posting semi-regular updates on their status
    • Communication issues between maintainer
      • Example: Nautilus UI changes in 3.5
      • Is it part of the release team remit to pester maintainers to communicate sweeping changes to a module in the release moduleset?
      • Another option: revert to a previous version until the regressions are fixed?
        • This already happens at the feature level (unmet features are pushed to the following cycle) and at the moduleset level (modules are pushed back or removed)
        • Examples: nautilus-cd-burner, gstreamer-1.0
    • Is the release team also in charge of the project's direction?
      • It would mean having the input or the presence of the design team inside the release team decision making process
    • Traditionally, the release team's role was to capture consensus in the community
      • The community sets the direction, and the r-t resolves conflicts between modules
      • But what if the community thinks that the r-t sets the direction?
    • Proposal: use GUADEC to set the direction of the project
      • Every other GUADEC sets the roadmap for the following two years
      • Reflected also in the accepted talks
      • Issue: possible marginalization of the maintainers and community members that cannot attend
      • Prior art: MozCamp

      • The issue of the direction of the community should be resolved within the community, through high bandwidth communication channels (e.g. GUADEC); after a period of public RFC, the release team is in charge of ensuring that the direction is followed through.
        • The release team would be the "editor" of the project direction, in terms on maintaining the modulesets that define GNOME
      • We need a benchmark to verify that the process is followed and it's working
        • Example of a benchmark for the Testable initiative: in the next round of outreach programs, all students should be able to build GNOME without going through jhbuild pains
    • Venues of communication
      • Go back using desktop-devel-list as a development mailing list, and require maintainers to join
      • desktop-devel-list has been used for multiple communication issues
      • It is difficult to know who the maintainers are
      • We could use the DOAP files and publish the current maintainers
      • ACTION: fredp to publish the list of maintainers extracted from the DOAP files on git.gnome.org

      • Proposal: use another mailing list for maintainers?
      • Issue: should it be public?
        • It should be up to the maintainers to decide
      • Issue: having yet another mailing list may not be well received
    • Enforcing the project's direction
      • The release team is already partly doing that by maintaining the moduleset
      • How to handle forks?
        • External forks are not under GNOME's purview
        • Internal forks should not happen
        • Maintainance releases from the old 2.x modulesets are not controversial and should not be considered forks
      • Case study: Nautilus in the 3.6 cycle
        • Definite lack of communication from the maintainers
        • Feature proposal period was not used
          • The rules were not clear for features involving a single project
        • Changes involved a set of feature removals
        • Proposal: the feature proposal period should also be used in case of changes involving a single project
          • Constraints: does it involve removing features? is the module part of core? is the module user visible?
          • The release team should exercise the option to punt the module out of the release moduleset until regressions are addressed or a sizeable discussion (determined by the release team) happened on the desktop-devel list
          • Some of the criteria could match the ones we use for the UI freeze requests, but the bar should be set higher
  • Inclusion of the Commit Digest on Planet GNOME
    • Requires amending the PGO policy
    • Not really controversial
    • Using a separate style to visually differentiate the feed on the Planet
    • ACTION: Emmanuele to contact the PGO editor to add the commit digest to the Planet, using a slightly modified style

New Action Items

  • ACTION: fredp to publish the list of maintainers extracted from the DOAP files on git.gnome.org

  • ACTION: Emmanuele to contact the PGO editor to add the commit digest to the Planet, using a slightly modified style

FoundationBoard/Minutes/IRC20120912 (last edited 2012-09-12 18:20:40 by EmmanueleBassi)