Search modal is triggered by clicking the search field on the navigation bar. After clicking, the search modal will pop up and user can type their search into the search box.
Results of a search are relevant to plugin name, description, author and keywords if there are any.
Results are in the format of a list of cards. If there is no relevant result, it will return a 'No result' string.
One small comment/question, how is trending determined? If done by download stats then we probably want to exclude those that are going to be bundled as a default plugin or they might always be at the top. (Apologies if this has been discussed or already actioned before).
In that way, the data fetched by the script must be persisted in Github or somewhere else. And we can use Github API to upload content to the repo. I'll try to implement it soon.
I believe we did, I think opinions were fairly strong on it. I was coming down on the side of "opt-in" anonymous telemetry but many were against it entirely.
that might goes now off-topic but isn't there any telemetry open-source telemetry product that would send data, which is clearly marked/recognizable within the Joplin code, and send it results to the github repo?
When the plugin website is coming online it might be worth trying to restart the discussion as that would improve UX.
In general, the problem with telemetry is that there are groups of people that will never enable it (e.g. power users, privacy-oriented people, enterprise users, etc.), so the data is skewed by default. Then you end up with statistics saying that users don't use any plugins at all, even though power users may have a lot of them installed.
This is why companies try to force telemetry any way possible, without even giving the user ability to disable it (e.g. what Microsoft does in Windows).
Statistics are all well and good, but I am totally against telemetry. If you are prompted at startup if you want to turn it on, ok. But I would not do it and as tomasz86 says the data are then also not really meaningful, because you get only inaccurate results.
I don't think including GitHub stars in determining trending plugins makes sense. It is more developer oriented than regular users. We might need to use Github API somewhere else though.
.
