Follow up question, what sources are the open runtimes based on? Is it similar to CentOS where the source code is compiled and the artifacts are distributed? Or is it more like Amazon, Microsoft and others have to reimplement the same features.
I’m guessing it all comes scale and bloat.
A hack from 5 years ago to store critical metadata in some schemaless blobs needs to be migrated from an unsupported system to a newer one. The new system uses an updated Java version, and it turns out this gets inexplicably slow when parsing parts of the metadata blobs. Also, maybe JNI is involved, or nobody knows how to build the project, etc.
If this was a startup then two devs would pull an all nighter and come up with an even worse hack. Or would offload that to a random cloud platform, burning vc money. Or would not use Java in the first place. Or the startup would not survive / pivot.
In a large corporation I guess there’s support.
They’re basically all based on openjdk source with different VMs and GC tech added in or supported to different levels.
I've lost count of how many times we had a problem, I went and read the proper documentation and said what had to be done, only to be ignored because upper mgmt didn't like the solution, so they called "support" only to be told the exact same thing.
Oracle uses a more tricky setup. Every LTS release and its updates, but only until 1 year after the next LTS release, is covered by their "no-fee license". After that 1 year the updates are under a different license that only allows personal/development use or use on Oracle cloud. So any commercial user needs to buy a license from Oracle for these versions.