Ambiguous definition of Micro URI with ID Remote

[gregmedd]

In the URI Spec, the micro encoding of a Remote ID URI is shown as containing two fields: an 8-bit ID_LEN followed by a variable length UAUTHORITY_ID. An annotation is provided on the UAUTHORITY_ID reading (1=256 bytes).

This (1=256 bytes) implies that the allowable size range for the ID is from 1 to 256 inclusive. Since the ID_LEN is 8-bits, it follows that its expected to store id.size() - 1 such that an ID with a single byte would be represented by a ID_LEN of 0. This is necessary if we read the allowable size range as inclusive of 256 as that value cannot be represented in an 8-bit unsigned integer.

However, Table 3 states that the requirements for ID_LEN are that it "MUST be greater than 0", implying that the actual range is 1 to 255 inclusive, with an ID_LEN value of 0 being an invalid value. That same table defines UAUTHORITY_ID as being "8-2040 bits" (2040 bits == 255 bytes), supporting this interpretation.

To remove the ambiguity, either the diagram in 4.2.2 should be updated to more clearly state the actual size range of UAUTHORITY_ID, or Table 3 should be updated to allow for up to 2048 bits in the ID while removing the language that a length of zero is disallowed.

[PLeVasseur]

The (1=256 bytes) is what's throwing you from this table, right?

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  UP_VERSION   |      TYPE     |           URESOURCE_ID        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|        UENTITY_ID             |  UE_VERSION   |   UNUSED      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    ID_LEN     |        UAUTHORITY_ID (1=256 bytes)  ...       |
|                                                               |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

I agree it conflicts with the other pieces of information in that section of the spec. My guess is that the UAUTHORITY_ID is intended to be [1, 255] in length.

In up-rust we comply with the spec: by having a set of validate_micro_form() functions for UAuthority, UEntity, UResource.

In particular the one for UAuthority checks the size is [1, 255].

[gregmedd]

Yes, that's the bit that is causing the conflict. I've been working to those diagrams, and from that standpoint it makes sense to use the full range of ID_LEN with the max UAUTHORITY_ID having 256 bytes. It's never valid to have zero bytes in the ID, so there isn't a reason to have a micro encoding that can represent an empty ID (if that were to change, we would presumably bump UP_VERSION).

However, if up-rust has already implemented this as [1, 255], then up-cpp can follow that. The spec should probably be updated so that table doesn't conflict with the earlier text.

[stevenhartley]

Hi @gregmedd you are right the limit is 1-255, can you push a fix to update the document (to correct this)? Thank you!!

[stevenhartley]

We are phasing out micro URI so this issue will be closed