Category: software

Subversion 1.11 is the first of the new 6-month regular releases with an emphasis on introducing new features more quickly and a shorter support period. For those requiring stability we have Long-Term Support releases; this is not one of those. For details see How We Plan Releases.

Changes I was involved with are:

• Improved Shelving (experimental)
• Commit checkpointing (experimental)
• Viewspec output command (experimental)
• Additions to svn info –show-item (‘schedule’ and ‘depth’)

Most interesting from my point of view is the already considerable enhancement of shelving compared with 1.10. No longer do we store the shelf as a patch file, but as complete before-and-after content, and we apply it in the same way ‘svn update’ or ‘svn merge’ would, except without (yet) any conflict handling.

The 1.11 version of shelving also adds a limited form of checkpointing, in the form of multiple versions of a shelf. It is analogous to how we might manually save “my-change-1.patch” and then a later version as “my-change-2.patch”. Shelving in Svn-1.11 summarizes the changes and gives an overview.

I should point out that while shelving is enhanced, it still has a substantial gap: it lacks support for copies and moves, and mkdir and rmdir. To fill this gap I am working now to put in place consistent internal APIs through which to push and pull changes into and out of the working copy storage layer. With those in place, an implementation of shelving that supports the full range of WC operations should be as simple as a glue layer that plugs a “pull from WC” API into a “push to shelf” API, and vice-versa.

The other notable improvement listed in the 1.11 release notes is one that Stefan Sperling has been devoting a lot of effort to:

• Improvements to the Conflict Resolver
  

That one is all about offering the user some sensible and useful options when tree conflicts occur.

Now I’ve got this release out the door I am looking forward to continuing development.

Lots of computer geeks love them, but I have an aversion to board games. At most once a year I can gently enjoy playing a round of Monopoly, draughts or even (if rather badly) chess. I trust the classics, knowing they’ve stood the test of time. What makes me cringe, though, is the board game that has been invented — no, that’s too kind — that has been produced as a brand tie-in, like The Top-Gear Board Game¹. Knowing the only motive is profit, how can I expect an enjoyable well-crafted game playing experience?

So what shall I make of C-Jump? It looks like the motive in this case is good educational intentions, and so instinctively I cringe for the game play. And I cringe for the C code. It claims to be “based on the code of a real computer program” but most of the moves are written as expressions with no assignment of the result (“x + 2;”).

At the same time, I am fascinated by the possibility it might be able to sow a seed of familiarity and fun in a child’s mind, just enough to provoke a curiosity later in life on encountering the same symbols again, in the same way that playing with fluffy toy horsies², even if they are green and blue and yellow, might for a certain child spark a vet’s career.

I have to admire the effort that seems to have been put into it. It might even be playable. Once a year.

---

¹ I have no idea if that’s really a Thing, but I bet it is. [UPDATE: Yes, I checked, it is.]

² Dear spelling checker, “horsies” does indeed have an “i” in it.

---

Vaguely related…

• a web site dedicated to Cycling Board Games