Ok, based on this I don’t think VT (virtual threads) related structure and performance improvements make sense. If you can’t do the other elements you’re forced to use a different implementation anyway and the performance issues can be addressed there.
Thanks for the feedback. Robert > On Jan 8, 2025, at 2:04 AM, Alan Bateman <alan.bate...@oracle.com> wrote: > > On 07/01/2025 15:18, robert engels wrote: >> Hi, >> >> I would like to revisit this. I have signed the OCA for another PR I worked >> on - I believe it covers all of my contributions. >> >> The websockets license looks to be very permissive. But it seems like this >> is something Oracle could reach out to the authors to resolve. > > The simple HTTP server in the JDK was never meant to be a fully featured and > high performance server. Instead it's meant for very simple usages, so think > getting started, the rite of passage to serve up a file, and local testing. > If someone wants a production ready server then there are dozens to choose > from. > > So I think highly questionable as to whether it's worth putting any time into > adding websockets or HTTP/2 support. > > >> I could also simply create a whiteroom ws impl, as I did for the http2 >> support - it’s a pretty small surface api. >> >> I am more than willing to work on this, but I need to know what the >> feasibility is of it actually being included. >> >> Even if the work was limited to “better support for vt”, some of the changes >> are more extensive (like the SSL handling). > There may be a few small things that would make sense to bring to net-dev to > discuss. I assume "vt" means virtual threads so if there is some issues there > then bring them to the mailing list to get some agreement that they are worth > doing. > > -Alan.
