Re[4]: [cgmo-webcgm] 5.7.10 Interface WebCGMEvent Example

[Benoit Bezaire <benoit@itedo.com>]

Hi Lofton,

Okay, well on my machine, with our latest implementation...

var cgmDoc; works
var cgmDoc = new Object(); works
var cgmDoc = new WebCGMMetafile(); does not work

The main problem in this example is Ecmascript variable scoping, which
I admit, I do not fully understand yet (need to read more about it).

I don't get how you could obtain:
"var cgmDoc = new WebCGMMetafile();"  -- worked.

Is there a fail safe somewhere that I don't see? Does IE handle that
case differently than ECMA-262, I don't know.

-- 
 Benoit   mailto:benoit@itedo.com

 
Tuesday, November 29, 2005, 12:45:22 PM, you wrote:

> Being short of time, I was doing trial and error to make the examples 
> (Event and EventListener) work.  Using WinXP, IE6, and Itedo Beta 5:

> "var cgmDoc;" -- did not work (JS object failure).

> "var cgmDoc = new Object();"  -- worked.

> "var cgmDoc = new WebCGMMetafile();"  -- worked.

> Now that I have a bit more time, I'm interested to hear the resolution of
> the questions you raise.  I'm not a JS/ES expert, but have to go back to my
> reference book to sort out this stuff.  Any guru out there have the answer
> for us?

> (We *do* need to sort it out, as it will need correction in W3C processing.)

> -Lofton.

> At 12:10 PM 11/29/2005 -0500, Benoit Bezaire wrote:
>>Hi Lofton,
>>
>> From my understanding of ECMA-262.
>>
>>var cgmDoc;
>>
>>Is an uninitialized variable (actually initialized to 'undefined');
>>
>>var cgmDoc = new WebCGMMetafile();
>>
>>Has an additional Initializer ( = new WebCGMMetafile();) the variable
>>is created (like above) when the scope is entered but assigned a
>>value when the variable statement is executed.
>>
>>In this case, because 'WebCGMMetafile' is not an known Object, the new
>>will throw a TypeError exception.
>>
>>You need to call getWebCGMMetafile(); to create a WebCGMMetafile
>>Object. Doing:
>>
>>var cgmDoc = new Object();
>>
>>gains you nothing (that I can see), since it will be re-assigned by
>>the getWebCGMMetafile(); call.
>>
>>That's my understanding, if someone has a different opinion, please
>>share it with the group.
>>
>>--
>>  Benoit   mailto:benoit@itedo.com
>>
>>
>>Tuesday, November 29, 2005, 11:39:04 AM, you wrote:
>>
>> > At 11:16 AM 11/29/2005 -0500, Benoit Bezaire wrote:
>> >><snip>
>> >>
>> >> >>   If other vendors are able to interpret such ecmascript, maybe they
>> >> >>   can say so.
>> >>
>> >> > Curious that you ask the question.  One reason I didn't catch the 
>> error is
>> >> > that your beta (b4 and b5) code executes both examples successfully, 
>> on my
>> >> > WinXP with IE6.  (Have you subsequently changed the code so that the
>> >> > invalid case fails?).  I *did* execute each example before 
>> submitting the
>> >> > text for CS ballot.
>> >>Are you sure you haven't changed var cgmDoc; to var cgmDoc = new
>> >>WebCGMMetafile(); between trying the test and publishing?
>>
>> > Well ... now I'm confused.  Both the listing in the CS text, and the code
>> > in the .html file (as shown by "View Source" after executing the html) say:
>>
>> > <script type="text/ecmascript">
>> >      var cgmDoc = new WebCGMMetafile();
>> >      function handleClick(evt) {
>> >      ....
>>
>> > Neither of them say "var cgmDoc;".  Your original question (previous
>> > message) was:
>>
>> >>  Is var cgmDoc = new WebCGMMetafile(); valid?
>>
>> > (And my answer was, "not valid, because the constructor function is not
>> > defined."  Now ... I don't know what JS or ECMAScript say about that case,
>> > if the constructor definition is missing.  Does it substitute "new
>> > Object()"?  Fail in unpredictable ways?  Other?)
>>
>> >>I just don't see how our implementation could create such an object.
>> >>I'll have to get back to you on this.
>>
>> > Does JS/ES language spec itself say anything about fallback or error
>> > mechanism for undefined constructor?
>>
>> > -Lofton.
>>
>>
>> >>--
>> >>  Benoit   mailto:benoit@itedo.com
>> >>
>> >>
>> >> > Another item for the issues/errata queue.  Hmmm... Dave, do you know 
>> how we
>> >> > handle errata pages in OASIS?  (It's not like W3C, where the errata 
>> page is
>> >> > linked off of the cover page "Status" section with a persistent 
>> URL.  But
>> >> > Mary mentioned it once and I think there is something like a convention
>> >> > with a fixed location/naming construct.)
>> >>
>> >> > -Lofton.