Atom Bidirectional Attribute
draft-snell-atompub-bidi-08

[Ballot discuss]
The document should specify what takes precedence when there is both unicode control character and a bidi attribute. Without this, there is no way for interoperability in the sense that the generator of a document can ensure that the receiver got the intended result.

The document should specify what takes precedence when there is html with internal bidi control indication along with an atom level bidi attribute.

[Ballot comment]
FYI, I would be uncomfortable standardizing this now, but I believe
this is an excellent experiment and look forward to results from the
experiment.

[Ballot discuss]
I found what I believe to be an accidental design error.  Here is a
proposed RFC editor note that would fix the error:

OLD:
  For compatibility with existing Atom documents that rely on direction
  guessing, user agents MAY perform direction guessing in documents
  that do not contain a "dir" attribute but they SHOULD NOT do so when
  an in-scope "dir" attribute is provided.
NEW:
  For compatibility with existing Atom documents that rely on direction
  guessing, user agents MAY perform direction guessing in documents
  that do not contain a "dir" attribute or contain a "dir" attribute
  with an empty-string value but they SHOULD NOT do so when
  an in-scope "dir" attribute with a non-empty value is provided.

James said: "As far as I can tell this specification is
complete.  It has received some feedback on the mailing list, which I
have answered and addressed in the draft.  There is at least one open
source implementation available (Apache Abdera).  Given that there are
not a lot of implementations out there, it would probably make sense to
look at an experimental RFC status but I will leave that up to you."