Strings referencing to OPC parts (2)

[Chris Rae]

Hi Murata-san - my apologies for sitting on this and a couple of other H/Z emails. Is there a chance we'd be able to schedule some time in Tokyo to talk through these as a group? I think your concerns cover an area that's not easily addressed by individual DR responses, so maybe if we can jointly determine as a working group what the best conceptual approach is, the solutions to the various DRs will write themselves.

Or, at least, be easier to write...

Chris

-----Original Message-----
From: MURATA Makoto (FAMILY Given) [mailto:eb2m-mrt at asahi-net.or.jp] 
Sent: 31 August 2010 10:06
To: e-SC34-WG4 at ecma-international.org
Subject: Strings referencing to OPC parts (2)

Dear colleagues,

H-identifiers are human friendly names for referencing to an OPC part within a given package.  An H-identifier cannot be used to locate a package.  In other words, an H-identifier is relative to the package.


1) The syntax of H-identifiers

I think that H-identifiers should be scheme-less (i.e., no "http", no
"pack", etc.) IRIs without fragment identifiers or queries.  Since,
LEIRIs are legacy and OPC is still new, I think that we should use
IRIs wherever possible, although the use of the cspace character would
be disallowed.

To be precise, an H-identifier shall be a non-empty absolute path in
an IRI.  See RFC 3987 for the definition of absolute paths.

Note: The syntax of non-empty absolute paths is defined by the
non-terminal "ipath-aboslute" in RFC 3987.

Note: By definition, an OPC part name may contain non-ASCII
characters, the space character, and any other characters listed in "5
Characters allowed in Legacy Extended IRIs but not in IRIs" in
http://www.w3.org/TR/leiri/.

However, I can live with the use of LEIRIs.  In other words,
H-identifiers are scheme-less LEIRIs without fragment identifiers or
queries.  See http://www.w3.org/TR/leiri/.

2) Conversion from H-identifiers to physical package item names

At present, the conversion is as follows:

Unicode string -> IRI -> URI -> logical package item name -> physical package item name

But it should be either.

LEIRI -> IRI -> URI -> logical package item name -> physical package item name

or 

IRI -> URI -> logical package item name -> physical package item name

I prefer the latter.  In the former, we can and should rely on
http://www.w3.org/TR/leiri/ for the definion of the conversion from 
LEIRIs to IRIs.  In the latter, nothing is needed.


3) Physical package item names

I think that we need some specification for standardizing physical
package item names.

At present, A.3 in Annex A provides a vague definition.

4) Conversion between logical packages item names and  physical package 
   item names (i.e., item names in ZIP files)

At present, A.3 in Annex A (non-normative) provides a vague
definition.

Cheers,
Makoto