Task description update
Added the below to "Open questions"

• 4. What – if any – editing interfaces, beyond the Reply Tool [and eventually, the New Discussion Tool] should Product Analytics and QA expect to see edit summary events emitted from?

I'm thinking feature: dtReply, action: toggle-advanced. It'll fire whenever you literally click on the toggle, but not when it's open when you initially view the page because the storage setting said it should be.

(dtReply for consistency with the logging added for the similar checkboxes in T254291.)

Change 627340 had a related patch set uploaded (by DLynch; owner: DLynch):
[mediawiki/extensions/DiscussionTools@master] Log when the advanced mode panel is toggled

https://gerrit.wikimedia.org/r/627340

Presumably we want to tell the difference between open and close?

The task description is ambiguous -- it just says to log when people engage with it. @ppelberg do we want to know that?

Eh, I've updated the patch to do it. It now does action advanced-show and advanced-hide.

Change 627340 merged by DLynch:
[mediawiki/extensions/DiscussionTools@master] Log when the advanced mode panel is toggled

https://gerrit.wikimedia.org/r/627340

In T261816#6460037, @Esanders wrote:

Presumably we want to tell the difference between open and close?

Good spot.

In T261816#6460090, @DLynch wrote:

Eh, I've updated the patch to do it. It now does action advanced-show and advanced-hide.

Sounds good. I've updated the task description's "Open questions" section with, what I understand to be, the events, their names and the schema within which they'll be logged.

I'm assigning this over to @MNeisler to review the above. Editing Engineering: please do not merge this patch until she's done so.

In the meantime, two resulting questions:

• @DLynch can you review the "Open questions" "1.," "2.," and "3." in the task description and correct any issues you see?
• @DLynch can you answer question "Open question" "4." in the task description?

It was merged before your comment, alas.

That said, 1, 2, and 3 all look correct.

4. What – if any – editing interfaces, beyond the Reply Tool [and eventually, the New Discussion Tool] should Product Analytics and QA expect to see edit summary events emitted from?

Off the top of my head, I don't think there's anything else analogous? We're not emitting events that are particularly tied to the concept of "editing the summary" -- we're emitting events that are explicitly about showing/hiding the advanced section of a feature called "dtReply". Edit summary isn't hidden away behind "advanced" generally, so it doesn't really come up.

In T261816#6460907, @DLynch wrote:

It was merged before your comment, alas.

Argh. This is on me. As a team, we'd previously agreed to explicitly indicating in the task description that a patch NOT be merged.

I think the deployment can go through in this case for the following reasons:

• We've previously decided where events of this kind should live (VEFeatureUse)
• The scope of the change is narrow (two events)
• The events themselves are relatively unambiguous and thus the naming of them is not likely to require much discussion
• The impact of instrumentation bugs is limited considering this is not an event we are immediately depending on for any decisions

That said, 1, 2, and 3 all look correct.

Okay, great.

In T261816#6460937, @DLynch wrote:

4. What – if any – editing interfaces, beyond the Reply Tool [and eventually, the New Discussion Tool] should Product Analytics and QA expect to see edit summary events emitted from?

Off the top of my head, I don't think there's anything else analogous?

Are you able to confirm this? I'd rather us spend the time now to be certain than try to untangle events when we're depending on the data meaning something specific.

Note: if you have conviction that there are no other events that would emit events similar to the ones we're implementing in this task and I'm misreading the "?" you included in T261816#6460937, please disregard the "confirm" question above.

Are you able to confirm this? I'd rather us spend the time now to be certain than try to untangle events when we're depending on the data meaning something specific.

Depends on the sense of the question, really.

They shouldn't expect that exact event combination to show up from anything else, because dtReply explicitly refers to the blob of UI that's sitting as part of the DiscussionTools reply widget, and only places that use that widget should use it. (I'm counting the new-topic tool as being equivalent here, since it's using the same widget.)

My uncertainty was over the advanced-show and advanced-hide actions, which could potentially show up attached to any other feature someday that has a similar panel that's toggled. There's nothing else semantically equivalent in the sense of being part of a save flow that's toggleable currently, at least. But it's easy to imagine us doing so on the mobile editor, say.

In T261816#6461021, @DLynch wrote:

Are you able to confirm this? I'd rather us spend the time now to be certain than try to untangle events when we're depending on the data meaning something specific.

Depends on the sense of the question, really...

Thank you for explaining this nuance, David. What you shared in T261816#6461021 makes sense to me [i].

i. Here is how I understand what you shared: the advanced-show/hide, in isolation could be emitted by other interfaces, but the specific dtReply string would not be.

I completed post-deployment data QA of the advanced affordance events. All data looks as expected. See observations below and notebook for further details:

• Events are being recorded as early as 17 September 2020, which was the date of deployment.
• Confirmed that both the advanced-show and advanced-hide events are only being recorded for dtreply feature and discussiontools integration on desktop as expected. Events appear for both editing interfaces.
• There are fewer advanced-hide events compared to advanced-show events as expected.
• As of 9 Oct 2020, there have been 85 advanced-show events and 47 advanced-show events (132 advanced affordance events total).
• As of 9 Oct 2020, advanced events recorded for the following arwikis: cswiki, enwiki, frwiki, hewiki, huwiki, jawiki, kowiki, nlwiki, svwiki, viwiki, and zhwiki.

@ppelberg - Let me know if you have any questions.

In T261816#6536823, @MNeisler wrote:

• As of 9 Oct 2020, advanced events recorded for the following arwikis: cswiki, enwiki, frwiki, hewiki, huwiki, jawiki, kowiki, nlwiki, svwiki, viwiki, and zhwiki.

@ppelberg - Let me know if you have any questions.

Thank you, @MNeisler. Is it surprising to you that no events were emitted at the Wikipedias* listed in the table below?

...I suspect "no" considering the low total volume of comments posted there.

*The Wikipedias listed in the table represent the Wikipedias where the Reply Tool was available between 17-Sep-2020 and 9-Oct-20202 and where no advanced-show and advanced-hide events were emitted. See: https://www.mediawiki.org/wiki/Talk_pages_project#Active_initiatives.
**One person is responsible for 98 of the comments posted at ka.wiki or 96%.

In T261816#6549152, @ppelberg wrote:

Thank you, @MNeisler. Is it surprising to you that no events were emitted at the Wikipedias* listed in the table below?

...I suspect "no" considering the low total volume of comments posted there.

@ppelberg
Given the number of comments posted on those wikis, I don't think it's surprising that we don't see any advanced affordance clicks. I rechecked the data today and it looks like there have been 2 affordance click events on kawiki since I last posted. Still no events for cawiki, ckbwiki or srwiki.

In T261816#6579789, @MNeisler wrote:

@ppelberg
Given the number of comments posted on those wikis, I don't think it's surprising that we don't see any advanced affordance clicks. I rechecked the data today and it looks like there have been 2 affordance click events on kawiki since I last posted. Still no events for cawiki, ckbwiki or srwiki.

Roger that, okay. Thank you for checking, Megan. Let's considered this done then.