Subscribing to commit notifications for the ET should be made easier

Issue #1422 resolved
Ian Hinder created an issue

At the moment, if you want to receive commit notification emails about changes to components of the ET, you have to subscribe to a number of separate mailing lists. There is a "commits" mailing list for the ET, but this currently only includes notifications for components which are not announced on other lists.

I propose that subscribing to the ET "commits" mailing list should be enough to receive commit notifications for all components in the toolkit, so that only one subscription is needed. This will also simplify mail filtering on the receiving end.

Note that the individual components could continue to manage their own mailing lists if needed, but a copy of the commit notification would be sent to the ET commits list.

The components with separate lists that I currently know about are:

  • SimFactory
  • Cactus
  • AEIThorns
  • Kranc
  • LSUThorns

Probably the easiest way to implement this would be to modify the VC commit hooks to send the notification to the ET commits list in addition to wherever it is currently being sent. The commits list may have to be configured to accept email from these additional addresses.

Keyword:

Comments (7)

  1. Ian Hinder reporter
    • removed comment

    I forgot that this was already discussed about a month ago. See http://lists.einsteintoolkit.org/pipermail/users/2013-July/003077.html:

    there is a desire to have fewer (ideally) one mailing lists for all commit messages or to be able to forward Cactus commits to the ET list. Problematic to allow users to Reply to these messages and have the answer forwarded to the original list. Shelved for now.

    I would like to open the discussion again. For me, the value of having a single source of commit messages outweighs the disadvantage of the reply-to header not pointing to the original mailing list.

    Having said that, it should be possible to configure the lists so that the reply-to header points to the appropriate mailing list. If the reply-to header is set in the VCS commit hook to the mailing list of the corresponding project, rather than in Mailman, and Mailman is configured to preserve the Reply-to header for the commits lists, we can still have the reply-to header of the commits mailing list point to the project-specific lists. i.e. you would have two commit hooks. e.g. for Cactus, one would send to the Cactus commits list, and the other to the ET commits list. Both would specify a reply-to header pointing to the Cactus users list.

    This assumes that the commit hooks are sophisticated enough to add such a header. I believe this is possible using svnnotify (http://search.cpan.org/dist/SVN-Notify/bin/svnnotify) and the multimail Git hook (https://github.com/git/git/tree/master/contrib/hooks/multimail). If using a repository host with its own hooks (BitBucket or GitHub, for example), this might not be possible. In that case, procmail could be used on the ET Mailman server to add a suitable reply-to header based on the corresponding project. Frank already indicated that he was interested in integrating spam filtering into Mailman, so this could be done at the same time.

    For reference, I believe that these are the current sources of commit notifications for all components of the ET. If we don't decide to send all commit notifications to the ET list (which I think we should, independent of the reply-to issue), then this information should be added to the ET webpage (http://einsteintoolkit.org/community/support/).

    || Repositories @ einsteintoolkit.org, Carpet, EinsteinExact, McLachlan || http://lists.einsteintoolkit.org/mailman/listinfo/commits || || Repositories @ cactuscode.org, GetComponents || http://cactuscode.org/mailman/listinfo/commits || || SimFactory || Email Erik Schnetter to be added (maybe also CCT support directly?) || || Outflow || "Watch" the GitHub repository https://github.com/rhaas80/Outflow (is there also an email hook?) || || Boost || "Watch" the GitHub repository https://github.com/dradice/Boost (is there also an email hook?) || || AEIThorns (5) || Email subversion at aei and ask to be added || || LSUThorns (4) || Email CCT IT support and ask for notifications (?) || || Kranc || "Watch" the GitHub repository (https://github.com/ianhinder/Kranc) or subscribe to the Kranc google group (kranc-commits) ||

  2. Frank Löffler
    • removed comment

    Figuring out the reply-to header should be possible, I agree. It is currently set anyway for that list. Ian: can you subscribe the ET commits mailing list to the Krank commits list and see what happens?

  3. Roland Haas
    • removed comment

    For github: they can send out emails like this:

    Branch: refs/heads/master
      Home:   https://github.com/rhaas80/Outflow
      Commit: 1dd3403f60e56e30ad67fe9959d58b435d72cd9a
          https://github.com/rhaas80/Outflow/commit/1dd3403f60e56e30ad67fe9959d58b435d72cd9a
      Author: Roland Haas <rhaas@tapir.caltech.edu>
      Date:   2013-08-12 (Mon, 12 Aug 2013)
    
      Changed paths:
        M README
    
      Log Message:
      -----------
      Outflow: update Tanja's email address
    

    this can be activated in the repository settings page under "Service Hooks" and then "Email". eg https://github.com/rhaas80/Outflow/settings#email_minibucket

  4. Ian Hinder reporter
    • removed comment

    I have enabled this, and (as per #1250) configured the From address to be the original committer. I have also tested the hook, but nothing has come through from the ET commits list yet. It's possible that the addresses need to be whitelisted in Mailman. As far as I can see, there is no way to configure the github commit hook to set the reply-to address to anything other than "no-reply@github.com". From comments by github developers, they "are trying to move people away from using email commit notifications". I'm not sure what would be better; email is a push notification receivable by anybody. Anyway, if Mailman could rewrite the reply-to header for these messages, this would solve the problem.

  5. Log in to comment