Conway's Game of Life (in pure T-SQL)

Have you ever wondered if you could create a simple game in SQL Server?

I did, for a long time, so one night I sat down and did it.

---

This is the result.

• [game of life](game of life.sql): this runs the game
• [pivot to grid](pivot to grid.sql): an sp which creates the board for the current generation
• [Life db script](Life db script.sql): run this to create the database itself - not necessary, there's nothing special about it

After this most recent spurt of key-smashing, the code is a little messy. Rather, the formatting and logical arrangement. There's a lot of clutter and comments. That will be addressed soon.

Performance

all tests are 100x100 board, 100 generations, SQL Server 2016 Express, Core i5-6500 3.2GHz, 8GB DDR4, SSD

"glider"	15s
"beacon"	27s
50 random coordinate pairs	about 30 seconds1	❓

There are further performance improvements that can be made. Right now I just limit my neighbor computation (which is the expensive part) to the rectangle of cells defined by the min/max row and column, plus a one cell border (because dead cells can have neighbors and this is important). This results in a huge performance increase for small, singular objects, but very little benefit for spread-out or large objects.

The other remaining difficulty is, how do we visualize this? I don't want to cheat and use something external to SSMS. AFAIK, there's no way to CTRL-R programatically in just T-SQL. Which is what I need. (That's not set in stone though.)

Anyways, it's working and I can't determine that there are any bugs. Enjoy!

This will continue to be updated as the project progresses.

Footnotes

1. The first generation took a very long time, then everything died, and it zipped through the 99 remaining generations. ↩