Analytics tools

[mirisuzanne]

Was talking with @aaronagray about the potential for Susy to track your usage, and report back. This would be limited to things we can actually track — nothing related to the actual DOM. A few things we could potentially track:

• How many times you call various Susy mixins
• How often you repeat certain requests (e.g. calling span(3 of 12) 16 times)
• How often Susy gives the same output (even based on different requests)
• Something related to similar selector chains? (We can access this with & now in Sass-script)

Would that be useful? What other ideas do you have? What would you want to see in a report? Discuss!

[scottkellum]

This is very cool. I would be most curious about the settings people are using and getting a better feel for how people are using things.

Compile time would be super useful and tracking that over releases. Is it going up or down?

[zellwk]

This would be cool! (2), (3) and (4) would be especially useful for me.

[mirisuzanne]

@scottkellum @zellwk interesting. Compile times would have to be tracked externally, but everything else here could be done in Sass. I love the idea of tracking how settings are used — and making some effort to collect that data (it would be a self-selecting sample, but still interesting). We could think about two audiences: the developer working on a project, and the broader Susy/Sass communities learning from what we've done.

[zellwk]

Definitely! Although Susy is great with the flexibility, we should probably start documenting down best practices and how to use her. It gets difficult for people to learn otherwise.

Insights into how a developer works on a project can be very very helpful indeed

[KittyGiraudel]

This should be done as a plugin and not part of the core I think. Also I am worried about the potential overhead it might introduce, especially in regard to performance.

Compile time would be super useful and tracking that over releases. Is it going up or down?

You will be facing the Observer Paradox here. To track compile time, you are loading extra code which are resulting in longer compile time. :)

[mirisuzanne]

Yeah, I think the data is interesting - but in the end it would be easier to collect manually with a survey.