Talk:Maze generation
Which algorithm? The linked wikipedia article has many. –Donal Fellows 13:24, 14 December 2010 (UTC)
- I am quite happy about the freedom to use any algorithm. I am just worried about every example being accompanied with a large maze. --Paddy3118 14:13, 14 December 2010 (UTC)
- Agreed. I also don't see the need to dictate a special algorithm. Its suitasbility might well depend on the language. --Abu 14:48, 14 December 2010 (UTC)
- You are right about the large examples. I reduced the size to just 6 lines. --Abu 14:58, 14 December 2010 (UTC)
- I recommend:
- Specifying a maze dimension, depending on output format. If it's going to be ASCII art, make it, e.g. 40x30. If it's going to be a raster image that can be embedded, make it 640x480. (Actually, for the ASCII art, I'd suggest using whatever the terminal size was for old 40-column-wide terminals.)
- Required that the algorithm used be identified, if possible. ("freestyle/homegrown" is an option, of course.)
- As/when the page gets excessively, split into per-algorithm subtasks, so the particular algorithms can be compared.
- As the subtasks again get large, split each examples' output into a separate page.
- If example code is large, break the code out to its own subpage, as is done with tasks like RCBF.
- I think that will allow the task to grow, and offers a reasonable balance of comparison, example freedom and page size along the way. --Michael Mol 15:31, 14 December 2010 (UTC)
- 40x30 is way too big. I've just changed the example output to 18x6.
- Also, specifying an algorithm is indeed better. I've changed the task to use the simple depth-first algorithm. --Abu 15:37, 14 December 2010 (UTC)