Closed agrueneberg closed 10 years ago
Let me answer my question myself: when a document is deleted in the current implementation, the parent folder will take on the most recent modification time from one of its children rather than creating a new one. By exposing the ETags of folders, this pull request might therefore mess up browser caches.
The spec just says "A successful DELETE request to a document MUST result in [...] the version of its parent folder being updated, as well as that of further ancestor folders." To me this looks like something we would have to fix in a separate pull request. What is your opinion?
Has this requirement been clarified at all in the spec? I don't want to add behaviour that's not explicitly mandated and required by clients.
@michielbdejong
@jcoglan it's better (more efficient) to add an ETag header. i thought about how to do this, and i think the easiest solution would be to generate one from the JSON string that is sent as the body - just take its SHA1. this takes a little bit of computing power on the server, but saves network and computing power on the client
draft-dejong-remotestorage-01 is not very explicit about returning ETags for folders, but it has a couple of paragraphs that can be interpreted that way, such as "All successful requests MUST return an 'ETag' header [HTTP] with, in the case of GET, the current version." Since you are storing and updating version information for folders anyway, why not expose them?