1. 6
  1.  

  2. 4

    I don’t think the release tag is appropriate here. Julian posted a design proposal earlier today. I didn’t even have time to read it yet. There’s a long road ahead of this being shipped in an official release. (@jcs, could I get an SVN developer hat?)

    1. 2

      (@jcs, could I get an SVN developer hat?)

      https://lobste.rs/hats/build_request

      1. 1

        My impression was that Subversion development had basically stalled, so this is really exciting to see, even if it’s only a firm and concrete proposal on how to implement it. Can I ask whether the proposal for explicit branches (I think it was called labels or something?) is also something y’all’re still thinking about? And any idea when a proposal like this might actually land, if accepted?

        1. 7

          Development hasn’t stalled, it has just become very slow. The project used to enjoy several full-time developer positions being funded by various companies, and that funding has now mostly vanished. A lot of developers used to regard Subversion as their regular day job, which makes switching the community to a fully volunteer model a bit difficult. Assembla is trying to reverse this trend, but so far they have only employed one developer. Let’s hope for the best :)

          The checkpointing and shelving issues in the bug tracker are quite old. In fact, they predate the migration of the issue tracker to the ASF’s Jira instance. So these are not new ideas. They’re being picked up by a company which has a business case that involves SVN and is funding a developer from the community to drive them forward.

          Explicit branches have been proposed over and over, but they are difficult due to the design and code that’s already there. Sometimes, it’s better to leave things that are “good enough” alone. I certainly had ambitions at some point. Julian’s (the same Julian) element-based merge was the most advanced attempt at this, but it was never completed. His current employer seems to prefer him to work on shelving and checkpointing, so it’s probably up to somebody else to finish his element-based merging work (more likely, I’d expect it will never get finished).