[Prev][Next][Index][Thread]
Re: Namespace Draft
>>Agreed, which is why the Depth header can take the value 0 (the collection
>itself), 1 (the immediate non-collection members of the collection),
>infinity (go as deep as you can). Do you see problems with this approach?
>
>That's not what the spec says s.v. 5.3; it lists only 1 and infinity,
>and I see I misread, thinking that values in between were allowed.
>I am confused by the terms "immediate member resources" (what is
>the force of "immediate") and "internal descendant resources"
>(not in 2. Terms).
Right. INDEX is defined to return a listing of the members of a
collection. In the example below:
A
/ \
J K
|
Z
Depth of 0 would be just A (and hence a very boring listing). Depth of 1
is A, J, K. Depth Infinity is A, J, K, Z. So, INDEX was limited to Depth
1 and Infinity so it would actually return something (Depth 0 would only
report information about the collection itself, perhaps its human-readable
title).
The intent of "immediate member resources" is the Depth 1 non-collection
resources, but the proposal should be more explicit in its language.
>In either case it would appear that External Member Resources
>would not be listed, yet their only sin is that they possess
>an absolute URI not relative to their parent's (the collection's?)
>URI. So INDEX will not give a complete listing of the contents
>of a collection if some are differently URI'd than the collection
>itself, even if they are on the same server. Is that right
>or am I missing something? (And did you intend to allow integers
>between 1 and infinity?)
Urrr, I think this is a problem with the wording of the proposal. The
intent is certainly to return a listing of external members during an
INDEX, but I agree with you, the proposal does not clearly indicate this is
the desired behavior.
>And what is a "noncollection member of the collection"? An
>internal member resource that is not itself a collection?
Yes. I agree, "noncollection member of the collection" is a really awkward
term.
- Jim