Some stress tests have a line like: //@ requireOptions("-e", "let iterations=1e5") if ["mips"].include?($architecture) to reduce the number of iterations on MIPS (since its bots are slower). This causes bugs with the bytecode cache enabled, seemingly because issues with (lack of) double quoting: stress/generator-cell-with-type.js.bytecode-cache: Could not open file: iterations=1e5 stress/generator-cell-with-type.js.bytecode-cache: Exception: ReferenceError: Can't find variable: let stress/generator-cell-with-type.js.bytecode-cache: global code@[Command Line]:1:4 This is a blocker to enable bytecode cache support on MIPS.
I'm pretty sure now this is just a generic bug in how parameters are dealt with in the bytecode cache helper script. Patch incoming.
Created attachment 459580 [details] v1
Comment on attachment 459580 [details] v1 r=me
Committed r294532 (250787@main): <https://commits.webkit.org/250787@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 459580 [details].