From: Lofton Henderson
[mailto:lofton@rockynet.com]
Sent: Tuesday, October 14, 2008 10:55 PM
To: mary.mcrae@oasis-open.org; 'Forrest Carpenter'; 'WebCGM'
Subject: RE: [cgmo-webcgm] style probery
At 07:18 PM 10/14/2008 -0400, Mary McRae wrote:
OASIS *does* have a namespace policy here:
http://docs.oasis-open.org/specGuidelines/namingGuidelines/resourceNaming.html
It is my belief that we meet these requirements, for the WebCGM
namespace. I have been using this document (of Robin's) for several
years, and in fact had discussions with him on this exact topic -- to what
should a namespace URI dereference?
The particular question of Forrest is about *private* namespaces for
application extensions to WebCGM's XML Companion File, in conformance with the
XML Companion File extension rules of WebCGM 2.0 Chapter 4.
Private means it is the property of Boeing, or Airbus, or SDI, or ... It
is not a normative OASIS namespace defined in an OASIS Standard.
I note that Forrest's question is, "what should be at the referenced
application namespace?" (For *private* namespaces corresponding to
application-specific extensions to XCF.)
Your document says, for an OASIS-specified namespace, "ideally
RDDL". That is not exactly a normative requirement, which is why the
WebCGM namespace has an informative dereference document there (I went though
all of this with Robin a couple years ago.)
One other question. What is the status of the policy document?
Early on, it says "this draft". (And following the link on
"approved" in the Status section, it quotes Robin, "[requested]
approval for trial implementation."
-Lofton.
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Tuesday, October 14, 2008 5:21 PM
To: Forrest Carpenter; 'WebCGM'
Subject: RE: [cgmo-webcgm] style probery
At 01:49 PM 10/14/2008 -0500, Forrest Carpenter wrote:
[...]
It after looking at the node-ns in the test suite it looks like the
get/setAttributeNS will work. I do not understand what if anything needs to be
at the namespace URI. If we use http://sdicgm.com
is that sufficient?
Have a look at the WebCGM namespace URI (sec 4.3):
http://www.cgmopen.org/schema/webcgm/
There are no firm rules (that I'm aware of) for what needs to be at a namespace
URI. Sometimes there is nothing (bad idea, IMHO). Sometimes
informal like WebCGM. Then there are some more formal proposals, like
RDDF. But no standards yet, AFAIK.
ACTION: we need to update that dereferenced file at the WebCGM namespace
URI, at least by the time of REC/OS.
-Lofton.
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Saturday, October 11, 2008 5:42 PM
To: WebCGM
Subject: RE: [cgmo-webcgm] style probery
All --
Reviewing this whole thread, it seems to me that get/setAttributeNS resolves
Forrest's need without introducing any additional functionality (ApsAttr, etc).
And it does so in the way intended for application extensions. To me, it
sounds like Forrest's use case is such an application extension. (I'm
still slightly unclear whether the particular application extension is
formulated as a private instanciation of 'screentip' functionality, which
Dieter correctly pointed out as something we try to avoid.)
It would be nice if we could agree on a resolution (such as this) in the
Wednesday telecon.
If "agreed", then a couple questions might come up:
1.) should the spec have some additional explanation/examples of these
features? (e.g. Forrest's could be an example);
2.) should the test suite have a test of get/setAttributeNS? (I think
"yes".)
Regards,
-Lofton.
At 11:30 AM 10/10/2008 -0400, Bezaire, Benoit wrote:
"urn:schemas-microsoft-com:vml" xmlns:o =
"urn:schemas-microsoft-com:office:office" xmlns:w =
"urn:schemas-microsoft-com:office:word">
Hi Forrest,
We have implemented get/set AttributeNS. I don't think a new API is needed
since get/set AttributeNS is meant exactly for use cases like the one you
describe. The code would look like this:
powerObj.setAttributeNS("http://acmeelectronics.com","acme:state",
"ON");
powerState = powerObj.getAttributeNS("http://acmeelectronics.com","state");
// will return "ON"
The application developper needs to come up with a namespace URI. In this case,
I came up with http://acmeelectronics.com
for example purposes.
Benoit.
From: Forrest Carpenter [mailto:forrest@sdicgm.com]
Sent: Friday, October 10, 2008 9:20 AM
To: Weidenbrueck, Dieter; 'WebCGM'
Subject: RE: [cgmo-webcgm] style probery
All,
What I am proposing is just a single metadata string, if we use get/set
AttributeNS will this be supported by other vendors?
Regards,
Forrest
From: Weidenbrueck, Dieter [mailto:dweidenbrueck@ptc.com]
Sent: Wednesday, October 08, 2008 4:49 PM
To: Forrest Carpenter; WebCGM
Subject: RE: [cgmo-webcgm] style probery
Forrest,
good clarification. A couple of comments:
We have a screentip attribute that could be used to display values like the
ones you describe.
If another attribute would exist that should be used to hold content for
screentips, it would be in conflict with the original screentip attribute.
If a NS attribute would be used, you could query the content via a script and
set the content as a screentip in one single event handler.
Anything that goes beyond this would require lots of definitions regarding the
meaning of that content string, the values stored therein, the behavior,
potential switches between the screentip attr and this one etc etc etc.
I would use AttrNS instead and do the rest in a script.
Regards,
Dieter
From: Forrest Carpenter [mailto:forrest@sdicgm.com]
Sent: Mittwoch, 8. Oktober 2008 23:42
To: 'WebCGM'
Subject: [cgmo-webcgm] style probery
All,
Clarification:
All a WebCGM viewer would be required would be to support the set/get the state
style property it would require no graphical effect from the WebCGM viewer.
An application that uses a WebCGM viewer could use this style property in
response to a mouse over event to display information about a grobject such as
voltage or to display a part number
We could easily use JS variables to accomplish the same thing, but to me the
code is simpler and more understandable if in response to a mouse over event
the application can get the state style property using the ID returned in the
mouse over function. If JS variables were used there would need to be a unique
event handler for each grobject or a long if, if else, function to match the ID
with the appropriate JS variable.
The get/set AttributeNS could be used, but I do not completely understand the
implications of this function. Are all compliant viewers required to support
anyone s name space? We have not implemented this function in our viewer and I
don t see any tests for this in the test suite.
If no one else sees any value in this other than SDI, I will be happy to
withdraw the request.
Regards,
Forrest