Expected behaviour
Visiting http://www.jneurosci.org/content/34/27/8976.long, as a member of a group that has annotated that doc, I should see annotations made there.
Actual behaviour
I don't.
Additional details
Consider this annotation created on that doc in the group:
{
"updated": "2016-02-03T22:16:57.450959+00:00",
"group": "5,..4",
"target": [{
"source": "http://www.jneurosci.org/content/34/27/8976.long"
}],
"uri": "http://www.jneurosci.org/content/34/27/8976.long",
"document": {
"title": ["Exercise Modulates Chloride Homeostasis after Spinal Cord Injury"]
},
"id": "AVKpNb1svTW_3w8Lytlo"
}
This API query finds the annotations (assuming use of a token belonging to a group member):
https://hypothes.is/api/search?limit=200&offset=0&uri=http://www.jneurosci.org/content/34/27/8976.long&_separate_replies=true
But this is the API query actually made by the H client:
https://hypothes.is/api/search?_separate_replies=true&group=5JPw23D4&limit=200&offset=0&order=asc&sort=created&uri=http%3A%2F%2Fjneurosci.org%2Fcontent%2F34%2F27%2F897
The client, with http://www.jneurosci.org/content/34/27/8976.long in its address bar, is looking up http://www.jneurosci.org/content/34/27/8976, not http://www.jneurosci.org/content/34/27/8976.long, and finding nothing.
Why?
I suspect because the doc contains:
<link rel="canonical" href="/content/34/27/8976" />
I'm aware that we made this server-side change https://github.com/hypothesis/h/blob/7f842ae3343df859cafa33a951d5e0508f0f304f/src/memex/storage.py#L227 to be more conservative about URI expansion. But that happened long before 2016-02-03. And in any case, this is a question about client behavior.
Here I am at http://www.jneurosci.org/content/34/27/8976.long posting an annotation, the uri in the payload is http://www.jneurosci.org/content/34/27/8976.
![image](https://cloud.githubusercontent.com/assets/46509/19581025/18ffb614-96df-11e6-9b61-5822690a8fd9.png)
Is that derived from <link rel="canonical" href="/content/34/27/8976" />
? Is that a change since the example annotation above was posted on 2016-02-03?
I suppose it's possible that jneurosci.org changed their rel="canonical" policy between then and now, but it seems unlikely, what they have now is what I'd expect. In which case, have we created a discontinuity? . If so, I guess resolving that would be a job for yet-to-be-created tools for viewing and perhaps modifying the doc alias table.