Re: Option #6: Using POST for uploading associated resources

[jpanzer@xxxxxxx (John Panzer)]

Janne Jalkanen wrote:

> > Cons:
> >
> > o It's not as full featured as WebDAV; it solves only the uploading 
> > problem; it doesn't allow for full control of a namespace by the 
> > client (but that might be a pro for the server side...)
> > o Doesn't solve the binary-content-in-XML problem at all (not a goal 
> > of this solution)
> > o Requires URL rewriting by the client
> > o If user really just wants to upload a picture with a little 
> > metadata, requires two uploads: this one, plus an Atom entry with 
> > some sort of "no, look over here!" link inside. 
>
>
> So, in order to upload, say three images I just took, and an entry, I 
> need to do seven transactions?
>
> 1) upload image #1
> 2) upload metadata for image #1
> 3) upload image #2
> 4) upload metadata for image #2
> 5) upload image #3
> 6) upload metadata for image #3
> 7) Combine results from steps 1-6, rewrite URLs in the entry, then 
> upload entry.

If this is a common case, I'd prefer something like this:

1) Upload image #1
2) Upload image #2
3) Upload image #3
4) Rewrite URLs in a referring entry
5) Post metadata for images #1,#2,#3 in a single Atom post, along with 
the entry

But, I'm not sure what the use case for this would be.  You have images, 
they each need metadata, _and_ you need to tie them together with an 
HTML entry as well?

> By the same logic, we would need (8) - upload metadata for the entry :)

No, the entry carries metadata without need to mention it. :)  That is, 
in the proposal, wherever you see "entry", read "entry with content, 
possibly (X)HTML, and standard Atom metadata".

> What if there was just a simple way to upload the metadata for the 
> binary objects, and the entry would still be separate?  Use just MIME 
> multipart to upload the image (essentially combining 1&2, 3&4, and 
> 5&6).  That would be somewhere half-way between the suggested "put 
> everything in a single, gigantic multipart message" and the Option 
> #6.  That would also mean that there would be no need for the server 
> to do any URL rewriting, as the entry would still be separate.

Some people seem very opposed to using MIME.  I'm not personally, but I 
think there might be objections to requiring its support. 

I really think that the common case is going to be:

(1) Upload a picture of a cat using POST, which itself has no metadata 
of its own.
(2) Post an Atom entry containing metadata _about the entry_, with HTML 
text like this: 
   Hey, look at my cat! <img src="..." align="right"/>

Not sure what you mean about URL rewriting.  There is always a need for 
URL rewriting, as long as the entry contains references to the binary 
objects.  See cat example above.

> > o Authentication for upload would have to be HTTP-based (Basic or 
> > Digest auth?) rather than something more XML-based, unless I'm 
> > missing something.
>
>
> WSSE should work well here.

Good point, it's just an HTTP header.

-John