Rewriting it. When I first created the program, I wasn't even sure it could be done realistically, and so I was just trying to get it to work. I have continued development on that initial program, but as the few of you know who have downloaded the source, it's really abysmal coding. To that end, I'm rewriting the entire thing from scratch (it's under 1000 lines of code, so it's not too bad), and trying to do it nicely this time. That being said, it will probably be a while until the next release, but when it is released, the source will be pushed on GitHub.
Output changes. I'm also making slight changes to the output. The program used to give an average value estimate simply by averaging out the values of the tile in each solution, but that's not accurate; you could get an average value of 2 with only solutions of 1 and 3. I'm planning to change it to display the odds of each value (along with the odds of voltorbs) to give the user a slightly better idea of what's going on. This will also give rise to a slightly more sophisticated tile-recommendation system, as I will now be able to recommend tiles based on how many possible solutions they exclude.
For people interested, I will be posting a few blog entries on some of the math and algorithms behind voltorb flip (including the absolutely beautiful constant-time removal of an element from an unsorted vector/array, which is usually linear-time)
Pending Updates
- A speed patch for after you move to a voltorb or a already-flipped tile
- Allowing for taking back moves
- Better error handling
- More cost-benefit analysis when recommending tiles
- A GUI
No comments:
Post a Comment