Version of pmap which uses JVM 21+ virtual threads when available, but will chunk the work up such that at most n concurrent virtual threads will be used (useful for workloads where system resource constraints could be exceeded e.g. maximum number of open file handles).

Notes:

• degrades to vanilla pmap on JVMs that don’t support virtual threads
• virtual thread version is partially lazy (results are computed eagerly, but merged lazily using concat)
• non virtual thread version ignores the n argument (since pmap chunks coll itself, based on the number of CPUs)
• each invocation of bounded-pmap* utilises an independent set of virtual threads, so parallel invocations may exceed system resource constraints

Version of future which uses JVM 21+ virtual threads when available.

Notes:

• degrades to vanilla future on JVMs that don’t support virtual threads

Version of future-call that uses JVM 21+ virtual threads when available.

Notes:

• degrades to vanilla future-call on JVMs that don’t support virtual threads

Version of pmap which uses JVM 21+ virtual threads when available, one per item in coll.

Notes:

• degrades to vanilla pmap on JVMs that don’t support virtual threads
• virtual thread version is not lazy

Are virtual threads in use on this JVM?