Organization of community Data Explorer queries

[nathankershaw]

We don’t seem to have a live library of Data Explorer queries here on meta.discourse.org at the moment. Would this be worthwhile addressing?

The topic referred to here is unstructured and has been closed as it was a mess:

@riking

Our community has got you covered!
Import as many queries as you like from here: What cool data explorer queries have you come up with?

@SidV has set up a great collection in GitHub:
discourse-data-explorer/querys.md at queries · SidVal/discourse-data-explorer · GitHub

Is the time right for a community managed library here on Meta? I think that many would find it useful.

[simon (Simon Cossar)]

@nathankershaw

The topic referred to here is unstructured and has been closed as it was a mess:

Yes, that topic got so many questions and responses that it became unreadable.

@nathankershaw

Is the time right for a community managed library here on Meta? I think that many would find it useful.

I think so. Does anyone have ideas about the best way to organize Data Explorer queries on this site? Ideally it would be possible to find existing queries by searching for a few keywords.

[justin (Justin DiRose)]

A few thoughts:

1. We could have a separate category with one query per topic. Could require approval from staff for new topics as to avoid clutter.
2. Could simply have a data-explorer-query tag for individual topics. These topics could reside at #howto:tips-and-tricks

[SidV]

@nathankershaw

(…) great collection in GitHub

Thanks for having me present

I’m open to changing the classification of the queries list on GitHub, maybe if we all think about a classification there, we can after debugging the categories well, go to something here in Meta (?).

Feel free to submit a PR

[nathankershaw]

@SidV

I’m open to changing the classification of the queries list on GitHub

I didn’t realise that each query can be PR’d, forked, etc etc. That is pretty cool, and I can see why you have done that. However, it is a confusing place for those who are less technically minded. All Discourse admins are familiar with Discourse, but only a proportion are familiar with GitHub.

What do you mean by classification? I could see tags working well for that side of things.

@justin

1. We could have a separate category with one query per topic. Could require approval from staff for new topics as to avoid clutter.
2. Could simply have a data-explorer-query tag for individual topics. These topics could reside at #howto:tips-and-tricks

I believe that a dedicated sub-category of #plugin would be best.

1. If we do go with one query per topic, there will be a lot of topics. @SidV has more than 70, and there are a lot more out there. This would polute / overwhelm #howto:tips-and-tricks significantly!
2. It would enable making every topic a Wiki, configuring go to first post, and having a suggested structure as a template
3. It would also mirror the configuration of the #plugin category, which is just nice!

[Johani (Joe)]

I propose moving all the helpful and efficient queries to one topic and closing it afterward.

This topic would be “the source of truth” for vetted queries.

We can then be a bit strict about the structure of entries that get added to that topic. Each entry would have a heading that describes it, a short blurb about what it does, and a mention to credit the author.

Once we do that, we can add a table of contents to that topic, and it would have a heading for each query.

To address Simon’s concern here.

@simon

Ideally it would be possible to find existing queries by searching for a few keywords.

The new version of the ToC component will have a real-time, in-post search bar that searches in the heading and the content of each section and looks like this.

(ignore the styles, for now, it’s still a WIP)

The new version will probably be production-ready around the end of the year.

As for the process of sharing/suggesting/requesting new queries, we can create a new topic for that. Good queries shared on that topic would then be added to the “the source of truth” topic that has the ToC.

[sam (Sam Saffron)]

I feel a far better enterprise would be figuring out a process for getting queries bundled in data explorer.

Perhaps a #feature tagged #data-explorer #candidate-query is created with the candidate query. The feature explains why it should be bundled and how it is unique and important.

We then decide if the proposal is valid, bundle the query and close it.

Then you can always browse through bundling request that got rejected if you really want.

[nathankershaw]

@sam

I feel a far better enterprise would be figuring out a process for getting queries bundled in data explorer.

Expanding the queries natively available in Data Explorer would certainly be very helpful and accessible for a lot of admins/users.

However, I believe there also exists a need for ‘second tier’ queries, and a structured place to discuss them. I come to meta looking for a solution to a problem I have, not to donate clever queries per se.

Perhaps @sam’s and @Johani’s approaches could be combined? A subcategory with some topics/queries tagged as ‘official’ which are bundled with the plugin.