Andrew Downes

Question: does this work in the browser

10

I appreciate this is a node library. Is there a version of this or anything similar that works in the browser?

Very long identifiers (data URIs)

4

We've recently come across an edge case where a system has used a [data URI](https://en.wikipedia.org/wiki/Data_URI_scheme) representing as an activity id. This appears to have occurred not by deliberate design, but...

Not rejecting extreme future timestamp statements to prevent issues due to clock errors is overkill

14

https://github.com/adlnet/xAPI-Spec/blob/master/xAPI-Data.md#247-timestamp > A Learning Record Provider MUST NOT use a future time for a "timestamp" property in a Statement. > An LRS SHOULD* NOT reject a timestamp for having a...

Example needs to be clearer it's talking about statement requests

8

>The LRS SHOULD* reject any request with 400 Bad Request status where the content type header does not match the content included in the request or where the structure of...

The word "Activity Object" in the requirement "Every value in the contextActivities Object MUST be either a single Activity Object or an array of Activity Objects." could link to: `#2441-when-the-objecttype-is-activity`

In xAPI we already have a concept of personas associated together as people. I've seen a few cases recently where an organization is looking at migrating to a new platform...

Can an LRS return an earlier Consistent Through time than it has returned to in a previous request?

1

Especially in the context of attachments, where statements-con-attachments are available later than statements-sans-attachments. See https://github.com/adlnet/lrs-conformance-test-suite/issues/207#issuecomment-314103768 and discussion there.

Update contributing.md link to correct branch to PR

1

See https://github.com/adlnet/xAPI-Spec/pull/1046#issuecomment-304072085

Clarify exactly which properties the LRS MUST (or SHOULD) omit when responding to a GET Statements request with format=ids

1

See https://github.com/adlnet/grail-issue-tracker/issues/61#issuecomment-272418796 I'm suggesting that the following is what's required of the LRS, mainly because this is what Watershed does today and it seems reasonable to me. - For agents...

"0 indicates return the maximum the server will allow." in the GET statements parametrs table in part 3: data. Should be LRS instead of server.