wilsonic

oh true

That's strange, I'll run some tests, thank you.

<a:1PepeRain:865419459241312277>

I sent it to you, thank you for your help.

sure no problem, I'm just going to show you what the problem is, thank you already ❤️

let me deploy here in product, 1 min

sure i will sendo to you and gary

I made the correction thanks, like this: ```typescript // before supabase.auth.onAuthStateChange(async (event, session) => { await disconnect(); // ⚠️ Chama supabase.removeChannel() }); // ✅ After supabase.auth.onAuthStateChange((event, session) => { setTimeout(async () => { // ← Wrapper em setTimeout await disconnect(); }, 0); }); ```

I don't know if it would be inconvenient, but if you want, I can upload a version with the proxy disabled for you to test and see what I'm talking about in practice

let me check 1 min

Yes, it was one of the first attempts. I thought it might be some extension affecting it, but I still had problems. I even tried using an anonymous guide

Okay thanks bro, I'm thinking that my only solution for now will be to use a proxy. I didn't want to have to do this, but it's a solution. Thank you very much. I'll wait to see if anyone has a solution to the problem. I'm willing to pay money if anyone knows how to solve it.

I don't know, is there anything I can show you to try and help me? Any logs? Any code? I greatly appreciate your help in responding to me. I've been trying to solve this problem for several months. The only solution I found was to implement the API via Vercel, and today I came across another solution: disabling HTTP/3 in Firefox. However, it's impossible to ask my users to disable it <:011blobcatworried:1203493457184690226>

Thanks Gary for the reply, great point about the async callback deadlock issue! I checked my code and I do have some async callbacks in onAuthStateChange. However, I don't think that's the root cause here because: Works perfectly in Chrome and Edge - if it were a deadlock, it would hang in all browsers, no ? Works when I disable HTTP/3 in Firefox (network.http.http3.enabled = false) - deadlocks are not protocol-dependent The timeout is exactly ~30s - typical deadlocks hang indefinitely, not at a specific timeout The consistent pattern is: Firefox + HTTP/3 + direct Supabase REST calls = 30s timeout That said, I'll fix the async callbacks anyway (wrapping them in setTimeout as recommended) to avoid potential issues. But I believe the core problem is related to how Firefox handles HTTP/3 (QUIC) connections that go idle. Given this behaviour, is there any known issue with Firefox plus HTTP/3 and Supabase REST, or any supported way to force HTTP/2 on the Supabase side so I can properly work around this?