On Mon, 30 Oct 2023 10:22:06 GMT, Johan Sjölen <jsjo...@openjdk.org> wrote:
> Ouch, I just realized that we can't differentiate between being provided with > the literal number 2 and `MemStatAction::print` anymore since you moved the > literal number parsing into this function. That means that we can't set > `print_final_memstat_report` after returning from this function. Yes, its a big annoyance with unnecessary templatization in this file. The natural point of "print_final.." would be in register_command, but I cannot (easily) query values there because value is T. So I cannot just do a "if type is uintx and option is Memstat and value is print then do xxx". This code would gain IMHO by being dumbed down. A multi-type value can just as well be expressed with a union, no templates needed. Especially since we keep the type info around at runtime anyway. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/16335#discussion_r1375991179
