Prototype

Comments and changes to this ticket

• 

  John-David Dalton August 8th, 2008 @ 04:59 PM

  • → State changed from “new” to “invalid”

  This should be fixed in the current trunk of Prototype.

• 

  John-David Dalton August 8th, 2008 @ 05:08 PM

  Ahh tricky, I see the bug now :)

• 

  John-David Dalton August 8th, 2008 @ 05:08 PM

  • → Milestone changed from “” to “1.6.0.3”
  • → State changed from “invalid” to “bug”
  • → Assigned user changed from “Mislav” to “John-David Dalton”
• 

  Mislav August 8th, 2008 @ 05:15 PM

  osmo tested with latest master. this still persists. can you tell why it's happening?

• 

  John-David Dalton August 8th, 2008 @ 05:19 PM

  Yes, I can see why. Its checking $(id) first to see if it finds one. If it doesnt then it checks to see if its a disconnected node. Well in this case since its a cloned node the ID exists in 2 places, one in the document, the other diconnected. So it matches the one in the dom and later when it does a check to see if its a descendant of the cloned node it will fail and return an empty array.

• 

  John-David Dalton August 8th, 2008 @ 05:20 PM

  I don't know if we should fix this because technically ID's are supposed to be unique. I wonder what the others think about this

• 

  John-David Dalton August 8th, 2008 @ 05:31 PM

  I could see a patch using somehting like: Element#isOrphaned()

• 

  Mislav August 8th, 2008 @ 05:33 PM

  I say wontfix because of non-unique IDs. That's what I first said to osmo on IRC.

• 

  John-David Dalton August 8th, 2008 @ 06:08 PM

  • → State changed from “bug” to “not_for_core”

  Andrew says won'tfix as well

• 

  osmo August 8th, 2008 @ 06:37 PM

  I agree that IDs have to be unique, but I disagree that IDs have to be unique beyond the Document (for example in different DocumentFragments, etc ..).

  e.g. I have a template in my Document. I use IDs in this template to mark important elements (elements that need to be updated or observed for example).

  
  var template = $('template_id').cloneNode(true);
  template.id = 'new_unique_id1';
  var template_content = template.select('#template_content_id');
  template_content.id = 'new_unique_id2';
  template_content.innerHTML = 'hello world!';
  
  // repeat x times
  
  document.body.appendChild(template);
  

  In what way is this bad practice or not conform with HTML standards?

• 

  John-David Dalton August 8th, 2008 @ 06:36 PM

  • → Milestone cleared.
  • → State changed from “not_for_core” to “bug”

  Ya I see what you are talking about (I will leave the ticket open so it will invite further discussion :)

• 

  John-David Dalton August 8th, 2008 @ 06:37 PM

  Can you try using css classes or custom attributes?

• 

  Juriy Zaytsev August 8th, 2008 @ 06:39 PM

  • → Milestone changed from “” to “1.6.0.3”

  This has nothing to do with "id" if I'm not mistaken. Here's an original issue: http://dev.rubyonrails.org/ticke...

• 

  osmo August 8th, 2008 @ 06:47 PM

  Thanks for re-opening the ticket!

  IDs are the more logical choice imo. I use classes for markup, and try to avoid using custom attributes (I do not want to go dojo style - the w3c validator does not recognise the custom attributes and invalidates the document).

• 

  John-David Dalton August 11th, 2008 @ 05:05 PM

  • no changes were found...

  • 0001-Fixed-an-issue-with-css-id-selectors-and-cloned-ele.patch 2.3 KB
• 

  John-David Dalton August 11th, 2008 @ 05:06 PM

  Juriy I think you are mistaken I explain whats happening in the comments above :)

  We can fix this by doing something like the attached patched/unit test.

• 

  John-David Dalton August 11th, 2008 @ 05:08 PM

  • → Assigned user changed from “John-David Dalton” to “Andrew Dupont”

  Also git trunk version didn't extend the resolved "targetNode"

• 

  Tobie Langel October 1st, 2008 @ 01:21 PM

  • → Tag changed from “element ie ie6 needs_patch needs_tests selector” to “element ie patched selector tested”
  • → Milestone cleared.