3529 - [RDF/XUL] ContentRemoved doesn't take care of a tree container's children

Status: VERIFIED FIXED

(Core Graveyard :: RDF, defect, P2)

Description

[Chris Waterson]

Can't "re-root" a xul:treeitem or xul:treebody by changing the value of the
'id' attribute.

Comment 1

[Chris Waterson]

Marked as M3, blocking Scott & mailnews from changing the folder pane. I am
working on this should have it done this afternoon.

Comment 2

[Chris Waterson]

Okay, I've implemented code in the generic builder that notices when the "id="
attribute changes on a piece of content and rebuilds the content model from
that point down. Hyatt: I'm reassigning this bug to you, because for all pieces
of content except the WidgetContentInsertionRoot, this is buggy. Specifically,
because of the "table hackery", the destruction of a parent node does not clean
up the children properly (this happens to work in XUL because I did a recursive
descent to destroy kids on the content side). We need to fix the tree code so
that it destroys child frames properly.

Anyway, Scott: you should be unblocked for dogfood because you just need to set
the "id=" attribute on the treebody anyway (which all works just fine and
dandy). Remember, use "treebody.setAttribute('id', newId)", _not_ "treebody.id
= newId", because we don't have JS reflection of attributes working yet.

Comment 3

[Chris Waterson]

Removed the M3 milestone, because the part that Scott needs is fixed now.

Comment 4

[Peter Trudelle]

setting p2 for m3

Comment 5

[chris hofmann]

Scott are you still blocked by this?  if no one is blocked lets move to M4.

Comment 6

[scottputterman]

I'm not sure yet.  As of Friday I wasn't able to get this to work(the error
could be my code), but I'll pull a new tree and try it out.

Comment 7

[David Hyatt]

This is identical to another bug already filed against me.  I'll mark the
other bug as a duplicate of this one.

Comment 8

[David Hyatt]

*** Bug 3426 has been marked as a duplicate of this bug. ***

Comment 9

[Chris Waterson]

Actually, this may be two separate bugs, but what the heck.

I think that I have a fix for the part where changing the ID of an element
screws up the content model (but not the part for destroying frames). But, I'm
not sure where the mail code lives that I'd use to test this theory. I have a
little test in my own tree that I've fixed...

Comment 10

[Scott MacGregor]

Chris, you can send me or scott Putterman a patch if you want us to try it out
for you. I'm about to send you email instructions with regards to running the
mail app. I'll be posting instructions to mailnews mozilla newsgroup later today
as well.

Comment 11

[David Hyatt]

My fix isn't required for M3.

Comment 12

[David Hyatt]

Pushing off to M5

Comment 13

[David Hyatt]

Fixed.  Waterson, you should let me know if the recursive descent hack over in
the builders is still there.  I couldn't find it.  YOu should only have to
remove the parent content node now... you don't have to descend into the
children and remove them now.

Comment 14

[gerardok]

Chris, can you help us verify this bug? If you agree the resolution is correct,
please mark it as Verified. Thanks!