58638 - DOM-CSS should support retrieval of shorthand properties

Status: RESOLVED DUPLICATE of bug 137688

(Core :: DOM: CSS Object Model, defect, P3)

Description

[buster]

using WinNT debug build, branch candidate 2 (with limbo bug fixes)
mouse over links in the "select category" box
get this assertion:

###!!! ASSERTION: can't query for shorthand properties: 'Error',
file S:\mozilla\layout\html\style\src\nsCSSDeclaration.cpp, line 3906

The result seems to be that links get highlighted but not de-hilighted.

Comment 1

[Pierre Saslawsky]

The assertions show that some DOM-CSS script using CSS shorthand properties is 
not supported. Reassigned to myself and updated the summary line (it was 
"assertion mousing over link").

For more info, see bug 49174 that I shouldn't have closed without first reading 
the spec at:
http://www.w3.org/TR/2000/CR-DOM-Level-2-20000307/css.html#CSS-CSS2Properties

Comment 2

[David Gardiner]

Newer W3C document:
http://www.w3.org/TR/2000/REC-DOM-Level-2-Style-20001113/css.html#CSS-
CSS2Properties

Comment 3

[Hixie (not reading bugmail)]

Netscape's standard compliance QA team reorganised itself once again, so taking 
remaining non-tables style bugs. Sorry about the spam. I tried to get this done 
directly at the database level, but apparently that is "not easy because of the 
shadow db", "plus it screws up the audit trail", so no can do...

Comment 4

[Pierre Saslawsky]

Reassigned to Daniel

Comment 5

[Christopher Hoess (gone)]

Adding keywords.

Comment 6

[Christopher Hoess (gone)]

Is this still valid? Which shorthand properties are we talking about, here? And
this is so in the wrong component...

Comment 7

[Christopher Hoess (gone)]

oops, reassigning.

Comment 8

[Pierre Saslawsky]

Christopher, please don't reassign a bug when you don't know what the bug is for. 
Instead, you can ask something like "shouldn't this bug be reassigned to jst?". 
Daniel was working on related problems lately.  Back to him.

Comment 9

[Anne (:annevk)]

Could someone make an attachment (I'll try to make one ASAP if nobody beated
me), I have the feeling this is fixed.

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

My patch for bug 125246 implemented retrieval of shorthand properties (which is
what this bug is about, I guess).  Daniel Glazman also did most of the work for
serialization to shorthand properties, which is a separate issue (although
there's still a bug open for the 'font' shorthand, and 'background' could use
some improvement).

There are still some issues with nsCSSDeclaration::GetValue(nsCSSProperty,
nsAString&), though, but they should probably be separate bugs.

A testcase would be nice, though...

Comment 11

[Anne (:annevk)]

Attachment: simple test case

Now I'm not sure if I understand it correctly, but if I did, this is the test
case we need.

Comment 12

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 137802 [details]
simple test case

No, this was about retrieval, not setting.

Comment 13

[Anne (:annevk)]

Attachment: simple test case

Let's try that again. If this is wrong again, I'm probably not the person to
write JS test cases...

Comment 14

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 137804 [details]
simple test case

This is far more specific than it should be.  You shouldn't rely on an exact
string being returned to test this bug.  (We probably return "0pt" instead of
"0".)

Comment 15

[Garrett Smith]

"border-color" is retrieved, but only when all four values are equal.
getComputedStyle(document.getElementById("footer"),'').getPropertyValue('borer-color');
Fails, returning the empty string.

getComputedStyle(document.getElementById("footer"),'').getPropertyValue('border-bottom-color')
Other values don't return anything. Padding: 10px - getComputedStyle(el,'').getPropertyValue('padding'); 

Returns the empty string.

This makes it very cumbersome to do javascript style transitions. 

Style transition:
1) read an element's style
2) make a transition to styles of a new object, over time.

{ padding: "100px", margin : "20px" }

It's step (1) that fails here. 

The expected result could be like what you get in IE:
javascript:alert(document.body.currentStyle.padding)

one of:
"0"; 
"10px 0 0"; // padding top is "10" padding left/bottom are "0"
"10px 0 0 11px"; // padding top is "10" padding left is "11"

IE gives the simplest string possible. 

Or it could be a non-shorthand value, always using all four values:
"0pt 0pt 0pt 0pt"

Comment 16

[Garrett Smith]

(In reply to comment #15)
> "border-color" is retrieved, but only when all four values are equal.

Sorry ,that's wrong. 

Comment 18

[Sebastian Zartner [:sebo]]

Wanted to add, that this would be needed for the Style Side Panel in Firebug.
See therefore http://code.google.com/p/fbug/issues/detail?id=1189.

Comment 19

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think this bug is a mix of:
 * duplicate of bug 125246 (fixed long ago)
 * duplicate of bug 137688
 * invalid (i.e., things that the spec requires)

If there are specific shorthands where we don't serialize them correctly even when we should, they should get bugs for each one.  (I believe layout/style/test/test_value_storage.html tests this for every shorthand value in layout/style/test/property_database.js, so we have pretty good test coverage.  In particular, see the tests that isnot(step1val, "", "...") and is(gDeclaration.getPropertyValue(property), step1val, "parse+serialize should be idempotent for ...").)

Comment 20

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(Also duplicate of the various bugs fixed while making test_value_storage.html's tests pass.)