Question about one of the old Autonomous Transaction approach

Hi All,

Regarding the discussion about Autonomous transaction in below message ID,
long time ago, it has been specified that having a new structure "Struct
PGAutonomousXACT" was rejected in PGCon hackers meeting. Can anyone know
why is it been rejected? What is the disadvantage/problem identified with
that approach?

CA+U5nMKEUm4abRQBndLYt5LEdekTAe8rbiYW3977YHMeOWQ1kA(at)mail(dot)gmail(dot)com
<https://www.postgresql.org/message-id/CA%2BU5nMKEUm4abRQBndLYt5LEdekTAe8rbiYW3977YHMeOWQ1kA%40mail.gmail.com>

I tried to look for answers going through various mails related to
Autonomous transaction with no luck. Any answer or hint about where to look
for answers will be helpful.

Thanks & Regards,
Vaishnavi,
Fujitsu Australia.