RE: DOM2 HTML / HTMLTableRowElement / rowIndex

>>>[Glenn Adams]
>>>I don't think you mean "logical" order, but, rather, "presentation or
display" order.

I agree that "logical" is not the best word choice. In any case, both IE
and Netscape use that order when indexing rows.

>>>Overall, I think that the current language is probably the correct
language

Whether logical, presentation, whatever, existing implementations do not
follow the current language of the spec, but rather this language: "When
determining rowIndex, The ordering of rows is such that THEAD rows are
first, followed by the rows of each TBODY section in document order,
followed lastly by TFOOT rows."

Respectfully,
Brad

-----Original Message-----
From: Glenn A. Adams [mailto:glenn@xfsi.com] 
Sent: Friday, September 20, 2002 8:41 PM
To: www-dom@w3.org
Cc: Brad Pettit
Subject: Re: DOM2 HTML / HTMLTableRowElement / rowIndex



>We would like to propose an erratum to DOM2 HTML regarding the rowIndex

>property of HTMLTableRowElement. This is so the DOM2 HTML specification

>will reflect the behavior of 2 currently interoperable implementations:

>IE 6 for Windows and Netscape 6.

>Where the specification reads: "This is in document tree order and not 
>display order. The rowIndex does not take into account sections (THEAD,

>TFOOT, or TBODY) within the table."

>The proposed erratum is: "This is in logical order and not in document 
>order. The rowIndex does take into account sections (THEAD, TFOOT, or
>TBODY) within the table, placing THEAD rows first in the index, 
>followed by TBODY rows, followed by TFOOT rows."

I don't think you mean "logical" order, but, rather, "presentation or
display" order. Most readers interpret document order as "logical"
order. Also, if you do base the definition of rowIndex on presentation
order, then you need to posit an indefinite height presentation
container (in a top-to-bottom block progression direction writing mode);
otherwise, page breaks that would cause repeats of header and footer
sections must be accounted for in determining the row corresponding to a
rowIndex; i.e., the same row may be assigned multiple rowIndices on
different pages. To pose it another way, in terms of your "logical"
model, do the footer rows at the end of the first page come after or
before the header rows repeated on the second page?

Overall, I think that the current language is probably the correct
language, and that the interpretation of rowIndex should *not* be based
on presentation order as this is dependent upon the formatting and
layout model being used, which is independent of DOM semantics, at least
at level 2.

>The documentation of the rowIndex parameter for 
>HTMLTableElement::deleteRow and HTMLTableElement::insertRow could be 
>clarified by language similar to the proposed erratum. For example, the

>HTMLTableElement::deleteRow specification currently states "This index 
>starts from 0 and is relative to all the rows contained inside the 
>table, regardless of section parentage." "Regardless of section 
>parentage" could mean order within the table element of the document, 
>or display order.

>Brad Pettit
>Microsoft Corp.

Regards,
Glenn Adams

Received on Saturday, 21 September 2002 01:48:02 UTC