30838 - complete implementation of nsRange interface

Status: RESOLVED WORKSFORME

(Core :: DOM: Core & HTML, defect, P3)

Description

[Kathleen :Brade]

Here is a snippet from an e-mail by Vidur:
Well...there are a few NS_ERROR_NOT_IMPLEMENTEDs in nsRange.cpp and the current 
version of the spec (http://www.w3.org/TR/DOM-Level-2/range.html) includes some 
new errors and exceptions. Last I heard from Joe, completing the implementation 
was not on his list.

I am creating this bug for tracking purposes but I don't expect Joe to get to it 
soon.  Maybe someone else can pick this up?  If not, hopefully it can wait until 
later in the development cycle.

Comment 1

[Kathleen :Brade]

M19 for now

Comment 2

[Joe Francis]

accepting bug

Comment 3

[ekrock's old account (dead)]

Marking FUTURE. Netscape can survive without complete nsRange for FCS.

Comment 4

[Dave...]

I'd like to suggest that at least nsRange::CloneContents/ExtractContents be 
implemented soon. It appears to be mostly complete already. 

Since the contents of a range cannot be extracted, currently the only usefulness 
of creating a range is the ability to delete it. 

The only way I can manipulate a range of the DOM right now is to use .innerHTML 
instead of ranges and then dump the content back into 
createContextualFragment(). 

Comment 5

[Dylan Schiemann]

This has been split into four bugs, one for each unimplemented method:
extractContents: http://bugzilla.mozilla.org/show_bug.cgi?id=58969
cloneContents: http://bugzilla.mozilla.org/show_bug.cgi?id=58970
insertNode: http://bugzilla.mozilla.org/show_bug.cgi?id=58972
surroundContents: http://bugzilla.mozilla.org/show_bug.cgi?id=58974

A limited testcase is posted for each.
All are assigned to [e-mail]

Comment 6

[Jeff Yates]

I am attaching a JavaScript "Patch" that can be used as a work around for the 
range object.  This patch adds support for the following:
    Range.innerHTML - read only
    Range.extractContents - as per W3C specs
    Range.cloneContents - as per W3C specs
    Range.insertNode - as per W3C specs
    Range.surroundContents - as per W3C specs
    Range.deleteContents - as per W3C specs - fixes buggy support from Mozilla
    Range.jmyCompareNode - extends Mozilla's compareNode to include
                           the following 2 constants
    Range.NODE_BEFORE_AND_INSIDE = -1;
    Range.NODE_INSIDE_AND_AFTER = -2;

I would like a C++ programmer to implement these in C if possible.

Jeff Yates

Comment 7

[Jeff Yates]

Attachment: Patch for Range functionality

Comment 8

[Jeff Yates]

Attachment: Updated patch with test cases and code commenting (had an error in surroundContent method)

Comment 9

[Jeff Yates]

Adding patch and test cases as text files since zip is not attaching properly.

Jeff

Comment 10

[Jeff Yates]

Attachment: JS Patch File

Comment 11

[Jeff Yates]

Attachment: Clone Contents Test Case

Comment 12

[Jeff Yates]

Attachment: Delete Contents Test Case

Comment 13

[Jeff Yates]

Attachment: Extract Contents Test Case

Comment 14

[Jeff Yates]

Attachment: Insert Node Test Case

Comment 15

[Jeff Yates]

Attachment: Surround Contents Test Case

Comment 16

[Jeff Yates]

The complete functionality of the Range object cannot be accomplished until 
ownerDocument bug (bug #27382) is fixed.  

As stated in bug 27382, the ownerDocument property of all orphaned nodes is 
null, thus the Range object cannot set container points within orphaned nodes 
(it generates a NS_ERROR_INVALID_POINTER).  This is against the purpose of the 
range object.  It is supposed to be able to navigate a DocumentFragment, which 
by definition is not attached to a document.

Jeff.

Comment 17

[Jeff Yates]

My final version of the Range Object Patch via JavaScript is now available.  
The first versions were concentrating upon HTML applications.  I have now 
corrected this oversite.  Since people have been having a problem with my 
attachments you can download the patch from http://pbwizard.com/Moz%
20Patches/RangePatch.zip.  Included in the zip file are the test cases that I 
used to test my script.

Jeff Yates.

Comment 18

[Jeff Yates]

Bad news.  I have just put in a bug (Bug #73552) that ALL of the comparison 
methods of the Range object fail if the range is the contents of an Attr 
object.  This is a major step back and my "Patch" does not fix this ... yet.

Jeff.

Comment 19

[Gervase Markham [:gerv]]

sicking: are you interested in converting this JS to C++? I know you've been
doing this sort of thing recently.

Gerv

Comment 20

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

I'll look into it after I'm done with the treeWalker (bug 82625). But wasn't 
most of nsRange fixed recently?

Comment 21

[Dylan Schiemann]

Yes, most of this was fixed.  This and 58969 seem to be duplicates now.

Comment 22

[Joe Francis]

more range for kin...

Comment 23

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Any remaining issues can be filed as separate bugs.