Comment 2 by jteh on 2008-05-30 02:45
This request is not as trivial as it might seem.

The most efficient way to navigate quickly is to find a node (element) of a given type; e.g. quick navigation for form fields, links, etc. We could move to the next node which isn't a link, but in a navigation bar, this could well be a list item which contains a link. This is obviously not what is desired. This means that we must resort to scanning the text of the buffer, which is not simple due to the complex nature of most documents. Also note that searching for the next piece of text which isn't contained within a link is not always very accurate; e.g. navigation bars containing unlinked items or headings.

There are more efficient, accurate ways of quickly navigating a page. For example:

• Using ctrl+downArrow will skip to the next block element.
• Some pages have links at the top which allow you to skip to the main page content.
• Many pages have headings at the start of the main content. Pressing h will jump to the next heading.

If anyone can come up with an algorithm (or even better, some code) to do this, we will definitely consider implementing it. However, at this stage, I don't believe this should block 0.6.
Changes:
Milestone changed from 0.6 to None

Comment 3 by Iris on 2008-06-12 07:02
Yes, that makes sense. I can see definitely how this would be hard to implement at the current time.
Sorry, I didn't see your response to this ticket and wasn't aware that this would be that hard to implement.

Comment 4 by jteh on 2009-07-04 04:34
I had a brainwave as to how this might be done. I'll see how I go with it.

Comment 5 by jteh on 2009-07-04 09:21
More than 5 hours, several attempts and a sore brain later, this is now implemented in r3037. In the end, it all boiled down to a rather elegant 18 lines of code, including comments.
Changes:
State: closed