295561 - [FIXr]DOM tables: can't access rows collections by ID

Status: VERIFIED FIXED

(Core :: DOM: Core & HTML, defect, P1)

Description

[Brett Donald]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.7.8) Gecko/20050511 Firefox/1.0.4
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.7.8) Gecko/20050511 Firefox/1.0.4

According to the DOM Level 2 spec I should be able to access the rows[] and
cells[] collections using either the ordinal or the id/name. However, I only
seem to be able to use an ordinal to access rows and cells.

This works: table_element.rows[0].cells[0]
This works in IE not FF: table_element.rows["row1"].cells["cell1"]
This works in IE not FF:
table_element.rows.namedItem("row1").cells.namedItem("cell1")


Reproducible: Always

Steps to Reproduce:
Browse to the supplied URL
Actual Results:  
No alert messages are displayed. JavaScript Console reports the following error:
Error: document.getElementById("table1").rows.namedItem("table1row2") has no
properties
Source File: http://castlehillsda.org/test_table.html
Line: 6

Expected Results:  
Two alert messages should fire when the page loads, displaying the contents of
the central cell in the table. This works correctly in IE6.

Comment 1

[Steve England [:stevee]]

Attachment: reporter's testcase from URL

Comment 2

[Martijn Wargers (dead)]

I think this is the spec you mean.
rows:
http://www.w3.org/TR/DOM-Level-2-HTML/html.html#ID-6156016
cells:
http://www.w3.org/TR/DOM-Level-2-HTML/html.html#ID-67349879
interface HTMLCollection:
http://www.w3.org/TR/DOM-Level-2-HTML/html.html#ID-75708506
The .namedItem method is also defined in the HTMLCollection:

This:
alert(document.getElementById("table1").rows["table1row2"].cells["table1row2cell2"]
is not correct.
This: rows["table1row2"] is a neat way in js and means practically the same as
rows.item("table1row2") and that is not correct because .item() method only
accepts an integer, see the interface HTMLCollection link.

This:
alert(document.getElementById("table1").rows.namedItem("table1row2").cells.namedItem("table1row2cell2").innerHTML)
seems correct though. I don't know why this one fails.

Comment 3

[Martijn Wargers (dead)]

Attachment: Testcase2

This works fine in Opera8 and IE6. And according to dom level2 html, this
should work, I think.
.namedItem() seems to work for form elements, though. See the testcase in bug
80627.

Comment 4

[Boris Zbarsky [:bzbarsky]]

The cells collection works fine; none of the testcases here actually test it
(they all die at the row collections part).

Martijn, in the DOM JS bindings HTMLCollection['string'] is mapped to
HTMLCollection.namedItem('string').  Using a numeric property id maps to item()
instead, as you pointed out.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Attachment: Patch

Comment 6

[Johnny Stenback (:jst)]

Comment on attachment 190981 [details] [diff] [review]
Patch

Nice :) r+sr=jst

Comment 7

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 190981 [details] [diff] [review]
Patch

Requesting 1.8b4 approval.  This is pretty safe and improves compat with both
spec and IE.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Fixed on trunk and 1.8 branch.

Comment 9

[Brett Donald]

Verified on:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b4) Gecko/20050817
Firefox/1.0+

Thanks zbarsky!

Comment 10

[Bernd]

the test case should be converted into a mochitest

Comment 11

[Aiko]

Attachment: MochiTest

Comment 12

[Aiko]

Checkin needed, please. I believe the Mochitest should be placed in mozilla/dom/tests/mochitest/bugs/ and Makefile.in must be modified?

Comment 13

[Boris Zbarsky [:bzbarsky]]

Actually, it needs to go in content/html/content/test.  And I'll expand on the test a bit before checking it in (test multiple tbodies, for example).

Oh, and in general you want the "checkin-needed" keyword, not a status whiteboard annotation.

Comment 14

[Aiko]

Thanks for the info. I was wondering about adding stuff to the testcase, too. For example what should happen if two or more rows have the same id (not allowed theoretically, but anyway)? First one wins, just like getElementById?

Comment 15

[Boris Zbarsky [:bzbarsky]]

Attachment: Test as diff

Comment 16

[Boris Zbarsky [:bzbarsky]]

Checked in the test

Comment 17

[Boris Zbarsky [:bzbarsky]]

Yeah, if there are multiple things with the same id, first one wins.

If you want to add more tests here, please do!