Implement a mechanism to generate and preserve a random id on the client side and transmit it to the JSONP interface each request.
|Resolved||None||T105868 Search gadget for subcategories and intersections search (2013)|
|Resolved||None||T105513 [DeepCat] Story: Tracking DeepCat gadget usage|
|Resolved||WMDE-Fisch||T109434 [DeepCat] Task: Generate, preserve and transmit random usage id for logging|
For the non Git-Hubers: A discussion about privacy is going on in the PR. I try to sum it up here.
Introducing the suggested changes might scare users, being tracked. We should be aware of that fact and avoid unnecessary data gathering. Since we do not know if the data we would get is of any use ( because just few people using the tool ) it might be useless logging it at all.
We were also discussing alternatives like also suggested here https://phabricator.wikimedia.org/T105513#1552602 to get usage data.
This is all true but still we wont know how the gadget gets used _if_ it is announced probably to a bigger audience. And having than usable results of how the community is using it would be really helpful to promote our goals. We can still focus on anonymity and privacy there keeping the data to the minimum.
About https://phabricator.wikimedia.org/T105513#1552602 it is a better approach. At least to know how many useres uses the gadget.
But it gives us now idea how the usage is spread across the users. ( power users vs. regular users ) and we can only make use of it, if the gadget is accepted to be added to that list. I doubt that this will happen a) in the near future and b) at all, if the functionality is still limited like it is now or will be if we don't find a solution with the Cirrus-Search team.
Everything we do should be as transparent as possible. If we merge this PR and roll the gadget with it the users will bin informed in detail on the official announcement page about the usage logging. We dont want to know which user searches what. We just want to have a brought idea of the general acceptance of the feature among users. And a vague idea how the users use the feature in concerns of frequency and intensity.