Sammy, In our implementation, we support the notion of posting compound entries using multipart/related. One part of the package contains the binary, the other part contains the Atom entry. This allows us to perform the post operation in a single step. I have it on my todo list to write it up as an extension draft.
- James Sammy wrote: >> ...why would you need to "update this metadata... > > Well, like if I wanted to POST an image with tags for example, I would have > to update the metadata right? Otherwise, just like you said, I'd have to > POST an Atom Entry instead of binary data and include the binary data in the > Atom Entry as a base64 inline resource. > > Are there any other options? I mean, why were you surprised that I would > want to POST an image with metadata? Sorry if I didn't understand your > comment, I just don't want to be missing something :-) > > Thank you, > Sammy > > "Thomas Broyer" <[EMAIL PROTECTED]> wrote in > message news:[EMAIL PROTECTED] >> 2006/4/28, Sammy <[EMAIL PROTECTED]>: >>> Just to clear a bit of confusion in my mind please: assuming a client >>> wants >>> to post a weblog entry with 3 images. The client is going to post each of >>> the images one by one, determine their readonly URIs from the server and >>> then post the actual weblog entry content. >> Yep. >> >> Or it might use a extension along the lines of PaceCompoundEntries. >> >>> 1- Doesn't this actually mean that the client is going to make 4 separate >>> POSTs to the server? >> Yes >> >>> 2- I'm assuming that the metadata of each image (for example: tags >>> associated with the image, exif data, etc.) is wrapped in an Atom Entry >>> associated with each image. So When I POST say "A picture of the beach" >>> per >>> the example given, I'll have to actually PUT again to the server URI >>> given >>> in the example (http://example.org/edit/first-post.atom) in order to >>> update >>> this metadata. Does this mean that I'm actually going to POST/PUT once >>> for >>> each image and then finally POST the Atom Entry representing the weblog? >> Er, why would you need to "update this metadata"? Otherwise, yes. >> >> Or you can POST Atom Entries with inlined base64-encoded image data, >> or use multipart/related (looks like PaceCompoundEntries but is not >> necessarily the same) or another multipart/* flavor, depending on what >> the server supports. >> >> FYI, there has also been discussion last fall about batch updates and >> eventually transactions. This won't be in the core protocol but might >> come as an extension as soon as someone has time to draft something >> ;-) >> >> -- >> Thomas Broyer >> >> > > > >
