tl;dr:
  Please display the whole designated "summary" message, at the top of
  the page, as if it were the first message in the conversation.  (And
  then again, in its proper place in the chronological order.)

It would be nice if there was a way to control (and amend) what shows
up at the start of a bug.  There's the existing "summary" feature is
very limited: it can only display a single, unformatted, paragraph.

Other bug trackers typically have a way to edit the main or first part
of the discussion.  I think we don't want that.

But, we could designate a particular message to show out of
chronological order, at the start of the web page.

That's what I thought the "summary" feature would do before I tried it
and then (when it didn't do what I wanted) read the documentation :-).
So ideally the feature I'm asking for here would be called "summary".

One option would be to simply change the current meaning of "summary"
and "outlook" to always display those messages first (and presumably
then again in their proper place in the discussion).  That's probably
better than some kind of transition where we have three our four such
features with different behaviours.

Ian.

On Thu, 11 Jun 2026, Ian Jackson wrote:n

This is partially because of implementation details; summary is a
control field and so it's easiest if it's an unformatted paragraph.[1]
[And partly to force people to really summarize the details of the bug
(the problem) and the summarize the outlook (how they're going to fix
it).]

One compromise would be to continue with the summary and outlook
paragraphs, but also track the specific message that set the summary or
the outlook so you could link to an "expanded" summary/outlook.

Another alternative would be to do away with the summary/outlook
paragraph entirely, and instead use them to pin a specific message to
the bug at the top (in addition to where it would go chronologically).


1: I'm not sure if we've defined how multiple paragraphs get parsed in
the control parser. I assume we'd ideally follow debian/control and use
periods for extension, but I'm not sure if the existing parsers actually
do.