I know I'm just preaching to the choir here but my favourite thing about open source/published source libraries/applications is discussions on bug reports/pr's like this.
It's just something so heartwarming of multiple people coming together to describe their symptoms, workarounds and theories of what could be causing it.
Wild hypothesising here on HN but if you read to the end of the GH issue users have been reporting that STUN has been failing (i.e. no P2P link establishment, fallback to high-latency relay servers.) Multiple users have been able to work around the issue by manually substituting older Valve WebRTC dlls. I'd love to read a postmortem from the Valve devs.
Halo MCC launched with virtually entirely broken networking because their networking stack couldn't do basic NAT hole-punching/traversal. And those folks gaslit 343 management (presumably, since Frankie and Bonnie more than covered for them), MS management, the internal testing folks, and then the public for months after launch. Two weeks+ of internal MS testing where everyone was like "uh it doesn't work" right before launch.
Granted, I expect a lot more out of Valve than the MS-culture/folks that led to 343i / Halo Studios (since from every single bit of evidence, the cultural problems continue).
The rabbit hole started as a major P2P issue in Israel and possibly other middle east countries and further investigations revealed it seems to be a worldwide problem.
interesting, people speculated that Street Fighter6 went from P2P to relay a few months ago on one of the updates. never wouldve thought it would be actually a valve issue
My unpopular opinion: Valve is basically a parasite or a landlord. They've been so successful it's hard to imagine a world without them, and they say "you gotta give the parasite its due" and we believe them and comply.
It's been kept around because they treat their customers ok, but they absolutely exsanguinate their developers.
And their engineering culture is... odd. They hire senior people and then let them all fuck sound aimlessly. Their APIs are terrible, their infrastructure is all over the place, they still have patch Tuesdays. But because they are the landlord that owns every house in town, what are you going to do, not pay rent?
Gabe is out there cruising the world in a billion dollar yacht, eating thousand dollar meals. All that came off the backs of developers who actually make the games.
> It's been kept around because they treat their customers ok, but they absolutely exsanguinate their developers.
This is true, but "treat their customers ok" goes a long way. When everybody else severely abuses their customers, the one company that doesn't generates a lot of goodwill.
Valve fascinates me because the devs there occasionally seem to be simply the best on earth in a given field, but despite that, bizarre bugs will persist for a long time. My favorite was how steam in home streaming from a PC to a steam deck wouldn't work if the steam deck had an Ethernet and wifi connection - one of the connections had to be disabled or the stream would always crash.
Maybe they need a few average devs there to spend time sweeping up behind the paragons that are pushing the envelope into these features existing at all.
Valve famously has a very flat org structure so it's possible that that problem just isn't sexy enough for someone to pick it up on their own, without being told by a higher-up.
I wish they offered remote; I'd happily work there doing those sorts of unglamorous bug fixes. High-reliability engineering is my jam.
Hm, I have always wanted to use this to play couch co-op remotely but is this even the same "service" that provides that?
Looks like they tracked it to a steam update in March, and there's a workaround for at lest 3 games that involves all players copying steamwebrtc.dll to the game's ./binaries folder.
Government-controlled inspection wouldn't be solved by switching to older DLLs (unless the code itself is compromised, which is unlikely for video game code)
It's just something so heartwarming of multiple people coming together to describe their symptoms, workarounds and theories of what could be causing it.
Granted, I expect a lot more out of Valve than the MS-culture/folks that led to 343i / Halo Studios (since from every single bit of evidence, the cultural problems continue).
It's been kept around because they treat their customers ok, but they absolutely exsanguinate their developers.
And their engineering culture is... odd. They hire senior people and then let them all fuck sound aimlessly. Their APIs are terrible, their infrastructure is all over the place, they still have patch Tuesdays. But because they are the landlord that owns every house in town, what are you going to do, not pay rent?
Gabe is out there cruising the world in a billion dollar yacht, eating thousand dollar meals. All that came off the backs of developers who actually make the games.
This is true, but "treat their customers ok" goes a long way. When everybody else severely abuses their customers, the one company that doesn't generates a lot of goodwill.
Maybe they need a few average devs there to spend time sweeping up behind the paragons that are pushing the envelope into these features existing at all.
I wish they offered remote; I'd happily work there doing those sorts of unglamorous bug fixes. High-reliability engineering is my jam.
Looks like they tracked it to a steam update in March, and there's a workaround for at lest 3 games that involves all players copying steamwebrtc.dll to the game's ./binaries folder.