Before you get too excited, keep two things in mind:<p>1) Using a single compression context for the whole stream means you have to keep that context active on the client and server while the connection is active. This may have a nontrivial memory cost, especially at high compression levels. (Don't set the compression window any larger than it needs to be!)<p>2) Using a single context also means that you can't decompress one frame without having read the whole stream that led up to that. This prevents some possible useful optimizations if you're "fanning out" messages to many recipients - if you're compressing each message individually, you can compress it once and send the same compressed message to every recipient.
Surely that is obvious to anyone who has compared zip and tgz? As long as you’re not affected by the drawbacks obviously.
Does streaming compression work if some packets are lost or arrive in a different order? Seems like the compression context may end up different on the encoding/decoding side.. or is that handled somehow?
WebSockets [1] run over TCP, and the messages are ordered.<p>There is RFC 9220 [2] that makes WebSockets go over QUIC (which is UDP-based). But that's still expected to expose a stream of bytes to the WebSocket, which still keeps the ordering guarantee.<p>[1]: <a href="https://datatracker.ietf.org/doc/html/rfc6455" rel="nofollow">https://datatracker.ietf.org/doc/html/rfc6455</a><p>[2]: <a href="https://datatracker.ietf.org/doc/rfc9220/" rel="nofollow">https://datatracker.ietf.org/doc/rfc9220/</a>
I think the underlying protocol would have to guarantee in order delivery - either via tcp (for http1, 2, or spdy), or in http3, within a single stream.
It sounds as though the data is being transferred over HTTP, so packet loss/reordering is all handled by TCP.
When I worked at Microsoft years ago, me and my team (a developer and a tester) built a high volume log collector.<p>We used a streaming compression format that was originally designed for IBM tape drives.<p>It was fast as hell and worked really well, and was gentle on CPU and it was easy to control memory usage.<p>In the early 2000s on a modest 2-proc AMD64 machine we ran out of fast Ethernet way before we felt CPU pressure.<p>We got hit by the SOAP mafia during Longhorn; we couldn’t convince the web services to adopt it; instead they made us enshittify our “2 bytes length, 2 bytes msgtype, structs-on-the-wire” speed demon with their XML crap.