RE: [office] ODF_1.0_Errata_4h - Adjustments - Removing 17.5

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

Michael, I agree with removal of any correction for the relative-path and
RFC-referencing paragraph.

I don't think it is necessary to delete the replacement for the "All other
kinds of IRI references ..." paragraph, but I will accept whatever version
that Patrick puts up on Sunday, October 19.

Here is an important reason for the "All other kinds of IRI references ..."
statement.  I don't believe it adds any restriction that is not already in
17.5.  If some of those restrictions are undesirable, 17.5 needs a different
kind of repair and/or deprecation or extension in ODF 1.2.

WE SHOULD KEEP THIS IN MIND FOR 1.2

RATIONALE:

   1. The last sentence of the original "All other kinds of IRI references
... " is "IRI
references inside a package may leave the package, but once they have
left the package, they never can return into the package or another one."
"Return ... into another one" would appear to be restrictive, depending how
one might interpret "return." - Is this only meant to prevent some kind of
mutual cross-reference?

   2. The observation about "special processing" is too much about what an
implementation might do and the resources that a host system might make
available.  The preceding paragraph (which is intact in this regard) makes
it clear that the effective base IRI must be realized by some means.  I
believe this is authoritative and the "special processing" observation is
both too specific and not necessary.  (They also use more terms that are not
used in the IRI specification itself.)

   3. The first paragraphs and the bulleted-item list at the beginning of
17.5 makes it clear to me that the counter-example is not permitted in
accordance with ODF 1.0:

"The following restrictions exist for IRIs that are used within a package:
...

"   * IRIs that reference a sub file of a package shall be relative, and
they shall not contain paths that are not within the package.  This
especially means that sub files of a package shall not be referenced by an
absolute IRI. [It does not restrict this to the same package in any way.
--dh:]

"   * sub file[s] of a package cannot be referenced from outside the
package, for instance from the file system or another package."

It seems to me that the change did not introduce a new issue but simply
reflect in a direct way the constraints that are already expressed in 17.5.
I do agree that the issue needs to be dealt with if this is an unintended
consequence of 17.5 as it is already written.

 - Dennis

-----Original Message-----
From: Michael.Brauer@Sun.COM [mailto:Michael.Brauer@Sun.COM] 
http://lists.oasis-open.org/archives/office/200810/msg00090.html
Sent: Friday, October 17, 2008 01:30
To: robert_weir@us.ibm.com
Cc: office@lists.oasis-open.org
Subject: Re: [office] ODF_1.0_Errata_4h - Adjustments

I agree. The best option we seem to have is to remove a clarification
for section 17.5 from this errata, and to resolve it in ODF 1.2. This
effects both clarifications, the one for the first paragraph, and the
one for the second. The reasons for this are below.

[ ... ]

 In particular, this [current paragraph] allowed to use an IRI like
"http://www.openoffice.org/docs/mydoc.odt/content.xml"; in an ODF 
document, where "http://www.openoffice.org/docs/mydoc.odt"; is a package. 
It only said that this is not resolved as expected.

The currently proposed text is:

***
Non-relative-path references shall not refer to files inside a package.
Relative-path references having paths that traverse out of the package 
shall not reference files inside any package.
***

Applied to above IRI, this means that a document that contains this IRI
is not a valid document any more, because this IRI references some file
within another package. We therefore would declare documents that were
conforming according to ODF 1.0 to be non-conforming. That seems to be a
substantial change.

[ ... ]