bump to plenary

punt to next week

bumped

Tess: looks like Yves had questions and they were answered...

Yves: ...

Tess: linked moz and webkit standards positions.. some interesting discussion. There is already a "beacon" spec - this is different. It's on purpose. But seems unfortunate that you have 2 similar sounding things with similar purposes. Feels like we should try to merge these things or find a way to integrate them.

Dan: that sounds architectural.

Tess: beyond that ... the basic issue

Dan: no user needs...

Tess: sites want to reliably know if something bad happened. But if something sufficiently bad happens then site can't reliably know. E.g. device crashes; browser crashes. There is a tension bwteen having a reliable signal and the reality of software in the wild.

Dan: state and the web being stateless...

Yves: you have local state with cookies -- local storage... so this is not about giving state.

Dan: can we push back and ask for some user needs?

Tess: right now the explainer only has dev needs - you can infer user needs. But I'd rather them to be explicit.

Peter: also there should be a user benefit aside from analytics and tracking.

Dan: leaves comment on user needs

Tess: will read the discussions and follow up.

Tess: Some feedback I've heard: insufficient integratoion with sendbeacon... Entirely new API that is in similar space to existing APIs... and maybe a better API shape would have been to add an argument to an existing API? Webkit's position -

Dan: Support but with concerns.

Yves: supposedly it's to save a state server side from the client. The fact that it can be done using GET is not really nice. Should be a post. Using a get is not ideal. will leave comment

Tess: Agree. I can leave a comment.

Yves noted that they're using HTTP GET inapproriately. It also looks like they're working on a very big change to this, so perhaps we should defer reviewing it until that happens.

Tess: ...linked to the agreement that we came to on their issue 70 ... there's actually a PR on the fetch spec... I thought we should close it. Fergal said the changes are in the API...

triaged

Sangwhan to post asking them to close, design has changed.

closed