Irssi over Tor glitchy and fragile. Am I doing it wrong?

cross-posted from: https://lemmy.sdf.org/post/52956979

It looks like the most common method to use irssi over tor is to use a transparent proxy to tamper with network libraries, like torsocks or using proxychains4. Those approaches are useless when you also use Irssi with #Bitlbee, because bitlbee runs a local agent obviously becomes unreachable with torsocks in the loop.

So I must use a more complex approach:

    
$ socat -T9999999 -s TCP4-LISTEN:13999,ignoreeof SOCKS4A:127.0.0.1:libera75jm6of4wxpxt4aynol3xjmbtxgfyjpu34ss4d7r7q2v5zrpyd.onion:6697,socksport=9050,ignoreeof &
$ socat_pid_libera=$!
$ irssi
$ kill ${socat_pid_libera}


  

Then irssi is configured to point the libera network to 127.0.0.1:13999.

That’s the idea. There is a separate socat process for every IRC host I might reach, which is about a dozen in my case. Apart from ugly tediousness, it works for like 30 min on avg then dies. I believe that’s the nature of Tor. Circuits die and get replaced, and when that happens socat is left with a dead connection for some reason.

Is there a remedy? I there a way to make socat resilient to tor volatility?