Currently the comment parser considers each heading to be the start of a new thread. If we were to change this it could potentially affect other behaviour, such as which users appear in the auto-complete list by default, e.g.:

==Topic==
 Comment by Ed
===Sub-topic ===
Comment by Peter

If I add a comment in "Sub-topic" it currently only considers Peter to be in that conversation.

This logic will also affect how we deal with notifications / watching threads. If we consider sub-headings to not start new topics, then does that mean you can't watch a sub-heading? Or if we consider the to be sub-topics, can you watch the parent and/or the sub-topic? Could that result in duplicate notifications, or confusion as to how much of the page you are watching?

We should look at other real-world situations (excluding "vote" because that isn't properly supported here) to see what is the likely expected behaviour.

As noted in today's meeting, the Reply Tool can now [i] recognize heading hierarchy which enables us to work on this task when it's prioritized.

i. T264478

By knowing the hierarchy we can choose which heading to use, but we still don't know which one is the most appropriate, for example this talk page:

== AfDs for 2021-03-14 ==
...
=== John Notfamous ===
...
==== For ====
Let's delete this article! [reply]
==== Against ====
...

If the user clicks reply, probably the most informative title would be the H3: "John Notfamous" as the title of the article being deleted, but the reply tool does not know if to use "For", "John Notfamous", or "AfDs for 2021-03-14". On many other pages the H2 would be more appropriate (as described in the task description).

In normal editing, we would go by which section-edit-link was used, but when the reply link is used we don't have that information.