[Edbrowse-dev] acid3 log!

[Edbrowse-dev] acid3 log!

[Kevin Carhart]

Til now I was not aware that there is a nice log variable being written to 
when you run the acidtests.  (Unless I found it and then forgot.)

We can echo it from jdb and find out some rich detail about the tests 
that are failing.

jdb
log

Test 00 failed: expected 'pre-wrap' but got 'undefined' - found unexpected 
computed style
Test 01 failed: undefined not callable
Test 02 failed: undefined not callable
Test 03 failed: undefined not callable
Test 04 failed: undefined not callable
Test 05 failed: undefined not callable
Test 06 failed: undefined not callable
Test 07 failed: undefined not callable
Test 08 failed: undefined not callable
Test 09 failed: undefined not callable
Test 11 failed: undefined not callable
Test 12 failed: undefined not callable
Test 13 failed: undefined not callable
Test 14 failed: no <iframe> support
Test 15 failed: no <iframe> support
Test 16 failed: cannot read property 'appendChild' of undefined
Test 17 failed: cannot read property 'hasAttribute' of undefined
Test 18 failed: expected '10' but got '1' - DOCTYPE nodeType wrong
Test 19 failed: expected '11' but got 'undefined' - document 
DOCUMENT_FRAGMENT_NODE constant missing or wrong
Test 20 failed: didn't raise the right exception for null byte in 
createElement()
Test 21 failed: undefined not callable
Test 22 failed: no exception for createElement('<div>')
Test 23 failed: expected '5' but got 'undefined' - wrong exception for 
createElementNS('null', '<div>')
Test 24 failed: expected 'update()  /* this attribute's value is tested in 
one of the tests */ ' but got 'function global() { [ecmascript code] }' - 
onload value wrong
Test 25 failed: exceptions don't have all the constants
Test 26 failed: undefined not callable
Test 27 failed: cannot read property 0 of null
Test 28 failed: getElementById() didn't return the right element
Test 29 failed: cannot read property 'cloneNode' of undefined
Test 30 failed: undefined not callable
Test 31 failed: undefined not callable
Test 32 failed: undefined not callable
Test 33 failed: expected '0' but got 'undefined' - failure 1
Test 34 failed: expected '0' but got 'undefined' - lang=english should not 
be matched by :lang(en)
Test 35 failed: expected '0' but got 'undefined' - root element, with no 
parent node, claims to be a :first-child
Test 36 failed: expected '0' but got 'undefined' - control test for 
:last-child failed
Test 37 failed: expected '0' but got 'undefined' - control test for 
:only-child failed
Test 38 failed: expected '1' but got 'undefined' - empty p element didn't 
match :empty
Test 39 failed: expected '1' but got 'undefined' - :nth-child(odd) failed 
with child 0
Test 40 failed: expected '1' but got 'undefined' - part 1:0
Test 41 failed: expected '0' but got 'undefined' - root was :not(:root)
Test 42 failed: expected '0' but got 'undefined' - failure 1
Test 43 failed: expected '0' but got 'undefined' - control failure
Test 44 failed: expected '0' but got 'undefined' - misparsed selectors
Test 45 failed: expected 'right' but got 'undefined' - body doesn't have 
floatation
Test 46 failed: cannot read property 'style' of undefined
Test 47 failed: expected 'auto' but got 'undefined' - control failed
Test 48 failed: cannot read property 'appendChild' of undefined
Test 49 failed: initially: tBodies
Test 50 failed: cannot read property 0 of undefined
Test 51 failed: undefined not callable
Test 52 failed: cannot read property 'elements' of undefined
Test 53 failed: value attribute wrong
Test 54 failed: cannot read property 'getAttribute' of undefined
Test 55 failed: cannot read property 'appendChild' of undefined
Test 56 failed: cannot read property 'appendChild' of undefined
Test 57 failed: undefined not callable
Test 58 failed: defaultSelected didn't take
Test 59 failed: expected 'submit' but got 'undefined' - <button> doesn't 
have type=submit
Test 60 failed: cannot read property 'setAttribute' of undefined
Test 61 failed: element had attribute on creation
Test 62 failed: cannot read property 'className' of undefined
Test 63 failed: cannot read property 'getAttribute' of undefined
Test 64 failed: expected './test.html' but got 'test.html' - object 
elements didn't resolve URIs correctly
Test 65 failed: cannot read property 'appendChild' of undefined
Test 69 failed: kungFuDeathGrip.title was null
Test 71 failed: missing document for test
Test 72 failed: missing document for test
Test 73 failed: cannot read property 'createElement' of undefined
Test 74 failed: contentDocument failed for <iframe> referencing an svg 
document.
Test 75 failed: contentDocument failed on <iframe> for svg document.
Test 76 failed: contentDocument failed on <object> for svg document.
Test 77 failed: contentDocument failed on <object> for svg document.
Test 78 failed: contentDocument failed on <object> for svg document.
Test 79 failed: cannot read property 'hasChildNodes' of undefined
Test 80 failed: linktest was null
Test 98 failed: undefined not callable
Total elapsed time: 0.17s

[Edbrowse-dev] acid3 log!

[Karl Dahlke]

Looks like we're batting 0 for 100.
If I wasn't horribly depressed about the rest of my life, then this would bother me.

Karl Dahlke

Re: [Edbrowse-dev] acid3 log!

[Kevin Carhart]

Hi Karl

Are you kidding, we just got iframe support thanks to you and several 
awesome people!  We were batting 23 for 100 before, and now I think a good 
10-20 of these tests are sure to be easy.  Each one is a nugget - they 
can be read and understood.  Most are no longer than the code for test 
zero that I emailed to the list. That's a lot easier to work with than a 
brick of minified crud.

Also, check out how many of those diagnostic messages are phrased alike:

Test 54 failed: cannot read property 'getAttribute' of undefined
Test 55 failed: cannot read property 'appendChild' of undefined
Test 56 failed: cannot read property 'appendChild' of undefined

If that is not three instances of the same issue, I will eat my hat.

It's because of iframe support that I am back in the acid3 file in the 
first place.  I'm sorry if sending 100 status messages all at once is 
overwhelming or got you down.

Call me optimistic but I think we're doing great!



On Tue, 15 Aug 2017, Karl Dahlke wrote:

> Looks like we're batting 0 for 100.
> If I wasn't horribly depressed about the rest of my life, then this would bother me.
>
> Karl Dahlke
> _______________________________________________
> Edbrowse-dev mailing list
> Edbrowse-dev@lists.the-brannons.com
> http://lists.the-brannons.com/mailman/listinfo/edbrowse-dev
>

--------
Kevin Carhart * 415 225 5306 * The Ten Ninety Nihilists

[Edbrowse-dev] acid3 log!

[Karl Dahlke]

Logical.

Karl Dahlke

[Edbrowse-dev] script in tree of document.written elements?

[Kevin Carhart]

I found out something potentially important and good.  It is relevant to 
test zero, and actually has implications for a lot of other tests.

Consider this mini page, hello.html

<HTML>
<BODY>
<SCRIPT>
document.write('<P>odd</P>');
</script>
</BODY>
</HTML>

Should the inserted <P> have <SCRIPT> in its series of parentNodes as it 
descends from document?

Here's my edited edbrowse transcript for this.  Asterisks signify my 
input.

* b hello.html
79
4
* jdb
* x1 = document.getElementsByTagName("P")
[object Object]
* x2 = x1[0]
[object Object]
* x2.parentNode.nodeName
script
* x2.parentNode.parentNode.nodeName
body

This is relevant to the acidtests because test zero runs removeChild of 
the last script.  The last script is this:

   <script type="text/javascript">document.write('<map name=""><area 
href="" shape="rect" coords="2,2,4,4" alt="<\'>"><iframe 
src="empty.png">FAIL<\/iframe><iframe 
src="empty.txt">FAIL<\/iframe><iframe src="empty.html" 
id="selectors"><\/iframe><form action="" name="form"><input 
type=HIDDEN><\/form><table><tr><td><p><\/tbody> 
<\/table><\/map>');</script>

So it merrily removes all of that stuff inside of document.write. 
Subsequent tests try to find those elements and they no longer exist!
If script does not occur in the tree of those iframe, form, etc,
then they will be available later on!

Kevin

[Edbrowse-dev] script in tree of document.written elements?

[Karl Dahlke]

[-- Attachment #1: Type: text/plain, Size: 817 bytes --]

It was not at all clear to me, is not at all clear to me, where to put the objects that result from document.write.
innerHTML is obvious, document.write isnot.
I thought they should parse under the script that does the document.write, that is reasonable, rational, but not based on any spec or anything.
If the document.write is from an onload or onclick function, not an obvious script,
then I put the objects in a nowhere place wherein they still get rendered so you see them in the buffer but aren't part of the js tree at all.
That can't be right.
See html.c line 559.
The short answer is, I could probably put them just about anywhere.
They could all go under document.body if that is where they're suppose to go.
Or document. You'll have to review the test and infer where they belong.

Karl Dahlke

Re: [Edbrowse-dev] script in tree of document.written elements?

[Kevin Carhart]

On Tue, 15 Aug 2017, Karl Dahlke wrote:

> It was not at all clear to me, is not at all clear to me, where to put the objects that result from document.write.
> innerHTML is obvious, document.write isnot.

Thank you for wrangling it in the first place.  I think it's 
semi-deprecated or Considered Harmful.  There are various articles 
frowning on it.  So it's one of those annoying constructs that shouldn't 
be used, but here it is in the tests itself.

Okay.  I just opened hello.html in firefox and firebug.  It puts the P 
immediately after the script.  The P has parentNode body.  <Script> is 
not part of its tree branch.

I'm also in acid3.acidtests.org in firebug right now to verify with this.
I'm opening up the document.forms.  There is only one and it is one of 
those tags that is part of the document.write inside of the last script, 
so I think it's an illustrative piece of evidence and it's easy to make 
sure you're on the element you want to be on.

document.form[0].parentNode is map, and
document.form[0].parentNode.parentNode is body

> I thought they should parse under the script that does the 
> document.write, that is reasonable, rational, but not based on any spec 
> or anything.

I think that's right.  And with script omitted from their descendancy 
trail/tree branch.  Could this actually be tidy-related?

This is going to be good.

Re: [Edbrowse-dev] script in tree of document.written elements?

[Kevin Carhart]

> Could this actually be tidy-related?

Actually, scratch that, since dw is javascript.

Re: [Edbrowse-dev] script in tree of document.written elements?

[Kevin Carhart]

On Wed, 16 Aug 2017, Karl Dahlke wrote:

> Ok, try it now - write objects under body.
> Only had to change a dozen lines of code.

That definitely did a lot!

There's a next thing going on, but there would be, wouldn't there.  Fixing 
one gets you the pleasure of being exposed to the problem that the first 
thing was masking out previously.  We can at least bask in the glory for 
a few minutes...