hovertemplate support for sankey

[antoinerg]

Closes #3269

[antoinerg]

I am not sure the description of hovertemplate attributes is satisfactory. I'd like to discuss the changes made in commit d952713.

Finally, because the inclusion hovertemplate in Fx.loneHover calls have been tested in other traces, is it necessary to test it again in sankey_test.js? Is it OK to only check for the existence of the expected variables in eventData? 🤔

[etpinard]

Hmm. Sankey event data is a bit of mess:

There's probably way more stuff it in than we need. Moreover, some of it isn't consistent with event data for other traces e.g. the trace key should be fullData. I'm not how much we can change it w/o making breaking changes.

Oh well, these descriptions look ok to me.

[antoinerg]

I added fullData to the eventData in commit 3204274

[etpinard]

v2 changes tracked in #168 (comment)

[etpinard]

Finally, because the inclusion hovertemplate in Fx.loneHover calls have been tested in other traces, is it necessary to test it again in sankey_test.js?

I would add one "true" hovertemplate test case where we check that a user-defined hovertemplate gives the correct hover label content. You're right though, no need to re-test all the loneHover logic again 👌

[antoinerg]

I would add one "true" hovertemplate test case where we check that a user-defined hovertemplate gives the correct hover label content.

Done in commit 3204274!

[etpinard]

@antoinerg oops I forgot about this PR.

Looks good 💃