This message was deleted.

Why do you not want it to be an

@OutputDirectory

? That sounds like the correct solution if you don't know the name before the task executes.

Complications with multiple instances of the task being wired to use the same output directory.

Can those instances not be merged, or use a separate directory each + one aggregator

Copy

task to merge the directories? Alternatively, is there anything preventing you from computing the name ahead of time, without writing the file?

Not computing the name, since that's randomized, but separate directories + aggregator is definitely doable.

I'm curious what the context is here. What requires a bunch of output files to be separately created by individual tasks, but also put them all in one output directory? That seems like it could have issues with the random names overlapping at some point.

Refactoring old Gradle code that did not declare their inputs and outputs correctly.