|
If you were logged in you would be able to see more operations.
|
|
|
| Proposal: |
In light of the atomic create/update issue, we should probably keep the stream as input to these two methods.
|
| Resolution: |
We will clarify in the protocols how the content stream should be encoded when provided in the create/checkin messages.
We will NOT add an option to retrieve the encoded content stream as part of getProperties, because it would be really inefficient. (It's equally inefficient on create, but in those cases we have an explicit requirement for including them to ensure compatibility with some existing repository limitations.)
We will clarify in the protocols how the content stream should be encoded when provided in the create/checkin messages.
We will NOT add an option to retrieve the encoded content stream as part of getProperties, because it would be really inefficient. (It's equally inefficient on create, but in those cases we have an explicit requirement for including them to ensure compatibility with some existing repository limitations.)
|
|
Sections: 3.4.1 & 3.7.3
1) Is this really necessary, as there is an entire set of stream services?
2) The schema (CMIS-Core) includes a type (cmisContentStreamType) but it is not referenced anywhere else except in CMIS-Messaging (which appears to be SOAP-related). How might this stream be incorporated into an Atom Entry (for REST) - is it Multi-Part?
3) It seems reasonable that a stream be provided to aid in atomic creates/updates.
4) If streams really are expected to be large enough to merit a sub-range requirement on getContentStream (see 3.4.7), then is it realistic to expect that one would base64 encode such a beast?
5) If the Entry document schema is modified to accept an encoded stream (or the spec clarified as to where this is placed within the Entry), then why would that stream also not be available from getProperties? This leads to an asymmetrical usage of streams.
6) For large streams where encoding is not practical, there should be a way to (for example) create a document which requires a stream, set that stream in a separate operation, and then commit/checkin that change (at which time the "required stream" is enforced).
|
|
Description
|
Sections: 3.4.1 & 3.7.3
1) Is this really necessary, as there is an entire set of stream services?
2) The schema (CMIS-Core) includes a type (cmisContentStreamType) but it is not referenced anywhere else except in CMIS-Messaging (which appears to be SOAP-related). How might this stream be incorporated into an Atom Entry (for REST) - is it Multi-Part?
3) It seems reasonable that a stream be provided to aid in atomic creates/updates.
4) If streams really are expected to be large enough to merit a sub-range requirement on getContentStream (see 3.4.7), then is it realistic to expect that one would base64 encode such a beast?
5) If the Entry document schema is modified to accept an encoded stream (or the spec clarified as to where this is placed within the Entry), then why would that stream also not be available from getProperties? This leads to an asymmetrical usage of streams.
6) For large streams where encoding is not practical, there should be a way to (for example) create a document which requires a stream, set that stream in a separate operation, and then commit/checkin that change (at which time the "required stream" is enforced). |
Show » |
|
Bug
Closed
Critical
createDocument & checkIn - Accept an optional contentStream input. How is this included and encoded?
In the REST binding, on create or checkin, the content can be specified in the atom entry content element either as part of the src attribute or contained in that tag. We should support that model since that is how APP functions