opcache after a year of property hooks

Two hundred and eighteen methods became properties, and the question of whether the opcode cache noticed.

  cached scripts       8,412 → 8,412   unchanged
  memory used          126 MB → 124 MB
  hit rate             99.4% → 99.4%

and the per-operation benchmark, unchanged from 2024:

  plain property       0.0021 µs
  asymmetric           0.0021 µs
  a get hook           0.0410 µs
  a method call        0.0398 µs

nothing moved. which is the answer, and it took ten
minutes to establish rather than being assumed.

A hook costs about what a method call costs because it is one, and the compiled representation is the same size. The reason to measure was that a year of conversions is exactly the sort of change that could have had a compounding cost nobody looked for — and it did not, which is worth one paragraph and no more.