278257 - Need a way to get the bound type of an instance node

Status: RESOLVED FIXED

(Core Graveyard :: XForms, defect)

Description

[Doron Rosenberg (IBM)]

In the MDG, for schema validation, I need to get at the value and type, and it
seems that getting the bound type is currently impossible.

<beaufour> I think I misuse the type attribues as an "any-type" thing in
GetNode() or something
<beaufour> and, the AddMIP()-calls in nsXFormsModelEngine needs to do specific
stuff for the type, including changing the AddMIP-call itself, and nsXFormsMDGNode()

Comment 1

[Doron Rosenberg (IBM)]

Note that we also need a way to get to the model element, as
nsXFormsModelElement has a 

nsCOMPtr<nsISchemaLoader> mSchemas;

which stores the external schemas the model references.

Comment 2

[Allan Beaufour]

Attachment: Support type and p3ptype in ProcessBind()

Changed nsXFormsModelElement::ProcessBind(), to support type info. It now sets
the @type and @p3ptype attributes on the instance nodes, so they can be
retrieved by probing the nsIDOMElement.

It also dispatches the proper xforms-binding-exception events.

I also added a not about the use of eModel_type in nsXFormsMDGEngince, and
changed the parameter type of AddMIP to use the enum while I was at it.

Two couple of questions:
1) Instead of setting the attributes via nsIDOMElement, I should use
nsIContent::GetAttr() instead, should be faster and lead to more simple code.
But then I need to get the namespaceid, etc. What do you say?

2) We were not using the XPath expressions saved in the property, so I deleted
it. We could save it there and use it in EvaluateNodeBinding. Should we? (on
another bug then)

Comment 3

[Doron Rosenberg (IBM)]

Comment on attachment 171512 [details] [diff] [review]
Support type and p3ptype in ProcessBind()

Please add a comment that we shouldn't be setting the type on the instance data
more than once, that was a bit confusing.

Also per IRC, nsIContent should not be used because it may change.

Comment 4

[Doron Rosenberg (IBM)]

Attachment: add a ::ValidateNode to xforms:model.

Comment 5

[Allan Beaufour]

Attachment: Review comments incorporated

I incorporated Doron's comments, his ValidateNode() function, and removed the
nsIContent comment.

I also changed ProcessBind() to return nsresult. It should have done that a
long time ago...

Comment 6

[Allan Beaufour]

*** Bug 257180 has been marked as a duplicate of this bug. ***

Comment 7

[Darin Fisher]

Comment on attachment 171629 [details] [diff] [review]
Review comments incorporated


>+  /**
>+   * Pointer to next nsXFormsMDGNode with same nsIDOMNode, but different
>+   * MIP type (mType)*/

nit: add a space before the */


>diff -X patch-excludes -upN -U8 xforms.newbranch/nsXFormsModelElement.cpp >   }

>-
>+  
>   // 5. dispatch xforms-rebuild, xforms-recalculate, xforms-revalidate

nit: remove random whitespace addition


sr=darin

Comment 8

[Doron Rosenberg (IBM)]

allan - I tried applying this to the trunk (with the landed xforms branch) and
it doesn't apply - manually trying to fix it shows it relies on missing
variables for example.  Seems something changed (bug 265931 perhaps) that broke
this.

Comment 9

[aaronr]

sorry, I had 265931 in my list of stuff that was in the branch, but it was
probably an older version of the patch in the branch.  When I made the trunk I
probably grabbed the latest patch that didn't get sr'd until today.

Comment 10

[Allan Beaufour]

(In reply to comment #9)
> sorry, I had 265931 in my list of stuff that was in the branch, but it was
> probably an older version of the patch in the branch.  When I made the trunk I
> probably grabbed the latest patch that didn't get sr'd until today.

Something went wrong, that's correct. Nevertheless I've fixed it now, and
checked it in to trunk.