Implement document.getElementById

[ILyoan]

document.getElementById() needs to be implemented very efficiently. O(1) is is desirable

[jdm]

@ILyoan Are you working on this, or is this an open task?

[ILyoan]

@jdm Not working, just opened.

[jdm]

@mihneadb Sounds like you could jump on this if you're interested.

[tetsuharuohzeki]

May I try to implement this?

[mihneadb]

Sure, go for it. Let me know if you change your mind.

[tetsuharuohzeki]

Thank you!

2013/8/24 Mihnea Dobrescu-Balaur notifications@github.com

Sure, go for it. Let me know if you change your mind.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/740#issuecomment-23201566
.

saneyuki_s
saneyuki.s.snyk@gmail.com

[jdm]

This is needed for #841.

[tetsuharuohzeki]

I think the basic logic of this should be:

• it would use std::hashmap simply.
• it need to update hashmap:
  • on instert element subtree to document.
  • on remove element subtree from document.

But I'm worrying about these points.

• Insertion point of hashmap (I suspect Document would be nice.)
• Access to Document from insert/remove function.
  • I'm trying it from hubbub_html_parser::parse_html().

How do you think about these?

[jdm]

Yes, storing the map in the document sounds fine. I think you might be able to just use the add_to_doc function for updating it on insertion right now, which gets called after parse_html when the whole tree exists.

[bzbarsky]

Storing in the document is the way to go.

It's worth thinking about whether to store just the first element with the given ID or all elements (in tree order) with the given ID in the hashmap.

You definitely don't want to do this from the parser; you want to update the map when elements are actually inserted/removed into the document, which can happen without the parser being involved... and conversely the parser can create elements not going into the document (see innerHTML).

[tetsuharuohzeki]

Hi. My WIP is this stage: tetsuharuohzeki@8685a91
This WIP's next step would be:

• Observe adding ids via Element::setAttribute().
• Compatibility with other browsers.

Before go to next step, I have some questions...

This works when elements are appended to/removed from Document.
But this doesn't work that this cannot store id & node pairs to Document.idmap on parse....

To resolve this problem, I doubt that https://github.com/mozilla/servo/blob/d465abdb1c2162ca9eeb72f391ea4a721332500a/src/components/script/script_task.rs#L714 : This document is transmuted from HTMLDocument. It doesn't have any idmap field.

Do you have any good idea about to store id & node pairs to this document?

[jdm]

Note: the earlier pull requests implemented the fundamental parts, but we still lack support for multiple elements with the same id.