search

ActiveTriples / linked-data-fragments

Basic linked data fragments endpoint.
Creative Commons Zero v1.0 Universal
15 stars 0 forks source link

How to Implement Cache Requests #4

Open tpendragon opened 9 years ago

tpendragon commented 9 years ago

Should we just use HTTP headers? The downside would be most browsers would always request past the cache.

chrpr commented 9 years ago

I think HTTP headers are the way to go? I'm not sure browsers requesting past the cache is that much of an issue. Do we have "browser access" use cases? Even if we do, it would be a problem for caching 3rd party sites like id.loc, but if a Hydra implementation included it's own triplestore for this content, we could probably configure it to return 304 Unmodified for those requests?

scande3 commented 9 years ago

I'm unsure what this is referring to exactly if it could be expanded upon? (Or can ask others at the OR2015 meeting tonight on it if I am just missing the obvious). Although, from an uninformed opinion based, HTTP headers would make sense to use for me.

tpendragon commented 9 years ago

I think the big question in my mind is how should a client request PAST the cache?

chrpr commented 9 years ago

When you say "client" do you mean Hydra, or other generic clients? If the latter, do we have use cases for that? If the former, there's gotta be a workaround via some "special" contract, though that isn't really ideal. Maybe a question to run by Ruben?

chrpr commented 9 years ago

@terrellt Thinking about this a bit more. Perhaps a client requests past the cache by issuing a DeleteResourceOperation, and then re-requesting the Resource, which presumably then redirects to a CreateResourceOperation?

Or should the client potentially go to the original source directly, get a new representation, than do a Replace in the cache? That puts a lot of responsibility on the client, but also seems clean.

How frequently would a client like Oregon Digital want to reach past the cache like this?