[jdk19] RFR: 8288425: Footprint regression due MH creation when initializing StringConcatFactory

[Claes Redestad]

On Tue, 28 Jun 2022 12:07:13 GMT, Jim Laskey <jlaskey at openjdk.org> wrote:

> LGTM

Thanks. FWIW it's a clean backport. 

> Is it possible to set up a regression test?

We caught this automatically - albeit a bit later - using our automated startup and footprint tests. 

A functional pass-or-fail test is always a bit awkward when dealing with performance characteristics. We could consider some test that asserts that a Hello World doesn't spin up, say, any  `LambdaForm` classes dynamically, but there's nothing _wrong_ with using such things early. Future improvements could also mean we can do so for "free" (say by caching such classes in CDS or pre-generating everything in jlink).

-------------

PR: https://git.openjdk.org/jdk19/pull/83