Reactor/Linear: Don't expand more than 3 mentions in one message

[lerlacher-fm]

This creates a lot of spam, so let's not.

It would maybe be more elegant to make a "would you like to expand these" question box. I'll make that a Fryday task.

[wolfsage]

Hmmm, I don't think I like this...

I noticed the example where you hit a problem, you did roughly:

* https://linear.com/org/issue/TEAM-123/the-description-that-synergy-expands
* https://linear.com/org/issue/TEAM-456/the-description-that-synergy-expands
* https://linear.com/org/issue/TEAM-789/the-description-that-synergy-expands

And then synergy basically repeated your post with TEAM-123 the description that synergy expands with TEAM-123 being a link.

I don't think that behaviour is useful, because synergy is expanding what's already a URL that includes the issue ID and the description of the issue into a different formatted version of the same information.

I think instead synergy should not expand those kinds of urls. That is, if it's already well formed and contains the ticket description, don't bother expanding it...

That way, this kind of post still gets replied to with useful info:

Did:
* TIC-123
* TIC-345
* TIC-678

but we don't double up on https://linear.com/org/issues/FOO-132-here-is-some-details.

I would also have synergy continue to expand https://linear.com/org/issues/FOO-456 if the issue id doesn't contain a description after it..

[lerlacher-fm]

@wolfsage I'm not sure what you're proposing to be the exact criteria for when to expand and when not to expand, and I'm not sure it would be useful.

I would also have synergy continue to expand https://linear.com/org/issues/FOO-456 if the issue id doesn't contain a description after it..

How would you code this?

[rjbs]

I think @wolfsage was suggesting that we'd expand "without limit" if the message contained only ABC-123 form strings, but up to some limit if they were already URLs. Is that right, Matthew?

[wolfsage]

I think @wolfsage was suggesting that we'd expand "without limit" if the message contained only ABC-123 form strings, but up to some limit if they were already URLs. Is that right, Matthew?

Err, no, I think we just don't have a limit, and only expand IDs / URLs that don't include a description.

If I post:

ABC-123
https://linear.com/ABC-345
https://linear.com/ABC-567-this-is-a-problem

The first two would be expanded, the third would not.

[rjbs]

I don't have strong feelings here, but I went and looked at some examples of this stuff in a scrum channel and I think I agree with Matthew's preference. If you like, I will do the work to make it work this way, @lerlacher-fm

Shall I, or are you heart-set on the implemented behavior?

[lerlacher-fm]

I'm not quite satisfied.

If I post:

ABC-123
https://linear.com/ABC-345
https://linear.com/ABC-567-this-is-a-problem

The first two would be expanded, the third would not.

IMO that is weird behavior that would be surprising to most people; worse, it wouldn't work for my specific bug-bear because sometimes I will just post "ABC-123" rather than the full link.

Would you be happy with the behavior of "if there's more than 3 links, then instead of unfurling, post a <would you like me to unfurl these? yes/no> dialog"?

[wolfsage]

What if we had a per-user preference so your things were limited or not expanded but mine were?

[wolfsage]

( You could do the "expand yes / no" if you like but I wouldn't want that for my stuff I'd just want it to expand )

[lerlacher-fm]

@wolfsage @rjbs I did the simple thing of making it a user pref. Let me know if you like it.

[rjbs]

Two easy fixes. Thanks for making the changes.

[rjbs]

LGTM. One inline suggestion made.

[rjbs]

Suggested change

	return unless (@matches <= $max_matches);
	return unless @matches <= $max_matches;

(the parens are not idiomatic in our perl)