353519 - Fire"document:content-changed" event when the document significantly changes

Status: RESOLVED WONTFIX

(Core :: Disability Access APIs, enhancement)

Description

[Aaron Leventhal]

ATK has a "document:content-changed" which should be fired on a document when significant changes occur. The idea is that is that in some cases it's faster for the AT to recache the entire document than try to invalidate small pieces of the document.

It's hard to know when this is, so for now the bug should be open to discuss. All DOM changes look the same to us. We know what part of the subtree is changing, that's about it. We just need to make sure we're really doing something useful that's worth the coding effort involved.

Here are some ideas:
1. Calculate the number of DOM or accessible nodes that are changing, and the number in the document. If more than half are changing, fire this event.
2. Another idea would be to invent a constant such as kMinNodesChangedBeforeRecache, and fire the event when that # of nodes has changed.
3. Never fire the event, because it's always faster to refetch a subtree than the entire document.

I tend to go with #3, because I don't see how it's ever faster to refetch all the nodes vs. some of them. Also, it's expensive for Mozilla just to calculate the number of nodes changing for either #1 or #2. Finally, in comparison with correctness bugs, this is less important. It's an enhancement.

I'm looking to the AT developers here for guidance.

Comment 1

[David Bolter [:davidb] (NeedInfo me for attention)]

Aaron, this event is new to me. Where is it defined/documented?  The only hits I'm getting are in lsr (http://www.google.com/codesearch?hl=en&lr=&q=%22document%3Acontent-changed%22&btnG=Search).

There is some vague mention of it here: http://accessibility.freestandards.org/a11yspecs/atspi/adoc/ADOC_ATK.htm#events

Is it possible the intention was to have the source for the event be a subtree or is it the entire document?

Regardless, I tend to agree with you on #3.

Comment 2

[Aaron Leventhal]

It's not well-defined. Let's leave it alone for now because of #3, but I don't want to close the issue right now. Let's wait until we fully support ATs and they definitely say they don't need it.

Comment 3

[David Bolter [:davidb] (NeedInfo me for attention)]

Mass un-assigning bugs assigned to Aaron.

Comment 4

[David Bolter [:davidb] (NeedInfo me for attention)]

Joanmarie, do you guys want this?

Comment 5

[alexander :surkov (:asurkov)]

I'm not sure it's needed, might be hard to implement.

Comment 6

[Fernando Herrera]

Atk has not such a signal, so we could not fire it.

I cannot find it either on at-spi nor at-spi2, the only reference is in that pyatspi constant.

Comment 7

[alexander :surkov (:asurkov)]

IA2 has (IA2_EVENT_DOCUMENT_CONTENT_CHANGED) but I don't think we should implement it.

Comment 8

[Marco Zehe (:MarcoZ)]

Closing based on comments #6 and #7.