[NLips]

Telecoms + networking software engineer here:

'tis true - the caller has call release control on the PSTN i.e. the call doesn't end until the caller puts down the phone. There's special handling for e.g. 911 calls so that call release control is given to the callee i.e. the PSAP.

This doesn't work for e.g. SIP because the SIP client is not a dumb slave to the network. If I click end call, even if the network doesn't 'want' to end the call, my phone thinks the call has ended.

[Rjevski]

This used to be true but I really doubt it’s the case anymore.

Most of the “PSTN” nowadays has a bunch of SIP or other digital stuff in the middle, so this breaks down. Not to mention, this was never possible for mobiles to begin with.

[NLips]

It really is the case. VOIP in the middle isn't a problem. MGCs can translate between the callee's phone being on-hook to a SIP re-INVITE with inactive media. In fact, I've worked on projects in the last few years to do just this.

Telcos move slowly.

You're right about mobile networks though - that's a different kettle of fish.

[Rjevski]

Just wondering, why is that? It seems like a lot of effort instead of just sending a BYE when the phone is hanged up. I don’t see any legitimate reason for this “feature” to exist - if anything it makes scams & eavesdropping by a malicious remote caller easier.

[jsjohnst]

Per one of the articles I found on it, British Telecom apparently set the delay to 3min. The rational (if article is to be trusted) is that customers wanted the ability to hang up one phone and pick up another phone in the same home without disconnecting the call.

[pwg]

It is how the old copper PSTN network worked, and if you are lucky enough to still have an actual copper PSTN phone line it is still true.

[Rjevski]

My argument was that even if you did have an “actual” copper line it would eventually be terminated at a device that speaks analog phone line on one end and spits out SIP on the other end.

This specific hanging-up behaviour was an artifact of older analog switches and I didn’t think they would emulate it in the software-based SIP switches but according to the comment above it’s still the case.

[jsjohnst]

I definitely knew this was the case back when analog COs were the norm, but didn't realize this was still true. Seems it is very dependent on PSTN provider (I know SWBell didn't do this for normal lines when I worked there) as some do, some don't.