-> http://www.w3.org/XML/XProc/2009/07/16-agenda
Accepted.
-> http://www.w3.org/XML/XProc/2009/07/02-minutes
Accepted.
No regrets heard.
Norm: We agreed to wait two weeks, I haven't seen any discussion. Any discussion now?
Proposed: Change the spec to state that the implicit output port binding does not apply to p:declare-step.
Accepted.
Norm: Also held over from two weeks ago.
... I haven't seen anything about this either, any discussion?
Proposed: Add the attribute.
Accepted.
Norm: What do we want to do in this case. For the record, Saxon at least, creates the path.
Vojtech: We don't create the path at the moment. We overwrite existing files, too.
Norm: I don't think I create the path either.
Alex: Why isn't this implementation defined?
Vojtech: If you can't create the path, then at the moment you can't create the directory.
Norm: As an implementor, I'm tempted to create the path, but I don't have any problem with making it implementation defined.
... Three choices: MUST NOT, MUST try, or implementation-defined.
Alex: I'm for implementation-defined.
Proposed: Make this explicitly implementation-defined.
Alex: Do we have an error code for this?
Norm: I think it would be err:XD0021.
Alex: Doesn't have permissions isn't the same as doesn't exist. So it might be handy.
Vojtech: There's also err:XC0050 in p:store which is specifically for the case where p:store couldn't store to that location.
Norm: Right.
Alex: So do we say when you should throw which?
Norm: I think we give implementors pretty blanket freedom to throw 21 whenever they want.
Any objections to the proposal to make it implementation-defined?
None heard. Accepted.
Norm: I don't think we want to do anything here, but I thought I'd raise it.
Some discussion of how messages may or may not be like stderr on p:exec.
Henry: So he could run the transformation by exec and grab stderr?
Norm: I suppose that would work.
Proposed: Close with no action.
Accepted.
Norm: Henry, any progress?
Henry: No, I've had other W3C priorities.
Norm: Ok. We'll leave it on the agenda.
Norm: It's important that we have a complete test suite if we want to get out of CR.
Vojtech: I made progress with serialization in my implementation, so I'll be writing new tests there soon.
Congratulations Vojtech on the release of your processor!
No other business heard.
Adjourned.