[edbrowse-dev] attributes

[edbrowse-dev] attributes

[Kevin Carhart]

I hope to try out Dominique's bootstrap problem today in quackit.

Here's a separate thing having to do with attributes.
It's great we got maerskline.com loading.  I went on to one of their 
applications, the track-a-container search from the homepage.


10
i1=maeu
11
i1*

There's a problem involving classList.  I tracked this down some dead ends 
and found out that it is coming from a routine in bundle.js called 
validate().

It wants to populate n, using getAttribute('id')

var t = e.parentNode, n =
document.querySelector('#'.concat(e.getAttribute('id'), '_error')), i =
!0;

But we return null, I believe.  And this is by design, isn't it?  I 
noticed that if I pass in id, implicitMember1 is called?  What does the 
implicit member code do?

As you can tell from the above JS, the page code is trying to 
e.getAttribute('id') and then this becomes part of the argument to 
querySelector.  n is supposed to get the resulting element, so when 
'null' is part of the argument to querySelector, there's no resulting 
element and it crashes a little later.

thanks, Kev

[edbrowse-dev] attributes

[Karl Dahlke]

[-- Attachment #1: Type: text/plain, Size: 660 bytes --]

I could use your help here.
Line 1680 is there because acid test 17 says that getAttribute of id should not return the id unless it was set by setAttribute, or that is the way I interpreted it at some point.
These various implicitMember functions are suppose to be exceptions, times when you don't getAttribute unless you did an earlier setAttribute.
Same for "class" acid test 61.
Maybe I'm not interpreting the acid tests right,
maybe I just need someone else to take a second look.

I have acid3 in the github source, it is my version of the original page, but I think it is true to that page, with my comments and changes of course.

Karl Dahlke

Re: [edbrowse-dev] attributes

[Kevin Carhart]

Hi Karl
Well sure enough.. I was unaware that there was any such distinction, to 
tell apart an attribute set at create time by createElement, from one that 
is explicitly set by setAttribute.  Ian the acid3 author has a comment 
right there on 61.  You assert that p.hasAttribute('class') must return 
false, and if it didn't, the diagnostic string says "element had attribute 
on creation."  I assume you're interpreting the acid tests correctly since 
you pored over the assert() conditions.  That would tell you what types of 
distinctions matter.

Maybe there is a JS problem in maersk where it is supposed to be setting 
id using setAttribute, earlier on.  That would be a plausible 
alternative hypothesis that threads the needle, and returning null would 
be correct.



  On Sat, 29 Jun 2019, Karl Dahlke wrote:

> I could use your help here.
> Line 1680 is there because acid test 17 says that getAttribute of id should not return the id unless it was set by setAttribute, or that is the way I interpreted it at some point.
> These various implicitMember functions are suppose to be exceptions, times when you don't getAttribute unless you did an earlier setAttribute.
> Same for "class" acid test 61.
> Maybe I'm not interpreting the acid tests right,
> maybe I just need someone else to take a second look.
>
> I have acid3 in the github source, it is my version of the original page, but I think it is true to that page, with my comments and changes of course.
>
> Karl Dahlke
>