retrigger-benchmark

⏱️ Benchmark run finished in 2m 13s
🧠 Peak memory usage: 5874 MB (↗ 0% more)

📈 Compared against baseline: 135s
🧮 Delta: -2s
🔢 Commit: 4f9f4f1

🟢 Generation time is unchanged.

You can retrigger the benchmark by commenting retrigger-benchmark.

LGTM
I will just improve the error handling in the metadata creation step by replacing .expect() with a more greaceful handling which otherwise would panic the program on failure, since the metadata saving isn't a crucial part of the generation pipeline. Let me know if you don't agree with that!

LGTM I will just improve the error handling in the metadata creation step by replacing .expect() with a more greaceful handling which otherwise would panic the program on failure, since the metadata saving isn't a crucial part of the generation pipeline. Let me know if you don't agree with that!

This probably isn't worth bike-shedding on two much, but my thinking is basically that it should be exceedingly rare for this step to fail but the other file saving steps (regions etc.) to succeed. Furthermore it may be confusing or frustrating for users if this step happens to fail and then, months down the line, they then need this file (like for one of the many arnis utilities that I expect to show up over time (: ). In that case it may be better to fail early, even if it requires them to regenerate their world

Like I said though, this isn't a huge deal either way and I don't feel strongly, just explaining my initial logic