23014 - percentage <iframe>s don't resize when window resizes

Status: VERIFIED FIXED

(Core :: Layout: Images, Video, and HTML Frames, defect, P3)

Description

[karnaze (gone)]

<iframe style="width:50%; height:300px" src=foo.html> doesn't resize properly
in all cases when the window resizes. It only seems to work when the window is
made smaller than the current width of the iframe.

Eric, you should follow the behavior of percentage <div>s rather than <table>s
because <table>s have a lot of legacy quirks. For example in the test case
above, if I had specified a 50% height, the <iframe> should not be 50% the
height of the viewport as is the case with tables.

Comment 1

[sam_bravard]

At least in M12, using HEIGHT=100%, WIDTH=100% the <IFRAME> doesn't size
properly at first (too small for the rendered window) then it only resizes if
you shrink the main window dimentions.

Personal rant: Far too many next gen sites and product UI's depend on this (not
to mention the one's we're building).  Since <LAYER> is dead, let's shoot for
getting IFRAME behavior as close to correct as possible in the first release
out the door.  Requesting a bump in priority.

Comment 2

[sam_bravard]

Backing status to P3 (you fellas own the priority, my apologies).  Putting my
votes behind this one.

Comment 3

[Peter 'Rufus' Nelson]

Not sure if it's a duplicate, but this is extreemly similar to <a
href="http://bugzilla.mozilla.org/show_bug.cgi?id=25453">bug #25453</a>

Comment 4

[Eric Pollmann]

Chris, can you help me out a bit understanding this one?  Since 50% height is 
not supposed to be 50% of the viewport, what should it be relative to?  Thanks!

Karnaze wrote:
> For example in the test case above, if I had specified a 50% height, the
> <iframe> should not be 50% the height of the viewport as is the case with

On the upside, I got % width working in my tree, and resizing the window will 
resize the iframe dynamically as it should, so I think I'm getting close on this 
one.

Comment 5

[Eric Pollmann]

*** Bug 25453 has been marked as a duplicate of this bug. ***

Comment 6

[Eric Pollmann]

I have a fix that I think does the right thing based on Chris Karnaze's comments
above, will aim for M16.

Comment 7

[Peter 'Rufus' Nelson]

*** Bug 35972 has been marked as a duplicate of this bug. ***

Comment 8

[Eric Pollmann]

After some help from Chris yesterday, I'm going to be making a few changes to my 
fix that will correct more cases than just percentage height and width.  
Specifically, we have problems if the height or width is unspecified, or if the 
iframe is in a table.  I'll try to get something sane to happen in all cases 
before checking this in.

Comment 9

[Eric Pollmann]

I have fixes for all cases except iframe percentage height.  It is unclear how 
this should / could be handled.  In every case where I specify a percentage 
height on an iframe (including when that iframe is inside a fixed height table), 
the reflow method gets an computed height and available height of 
NS_UNCONSTRAINED.  As such, we're currently treating it exactly like the height 
wasn't specified and giving the iframe a height of 150.

Consider this case:
<table border height=600>
 <tr>
  <td>
   <iframe height=50% src=about:blank></iframe>
  </td>
 </tr>
</table>

The page designer might expect an iframe of height 300.  However, since the 
iframe doesn't know that the table is going to be 600 pixels high (gets 
NS_UNCONSTRAINED for computed and available height), we just make it 150 pixels 
high.

Even worse, what if the table had a height specified of 40 pixels!  Then the 
iframe being 150 forces the table to a height of 150.  Not sure what to do here.

Comment 10

[Eric Pollmann]

Chris, your suggested change to ReflowState worked!  :)  (making iframes 
caculate percentage height like tables in quirks mode)  I'll run through all the 
test cases and pre-checkins.

Comment 11

[Eric Pollmann]

Percent width fixes checked in.  I have percent height fixes in my tree, which 
need to be reviewed.  I'll attach them shortly (and send them your way Chris)  
:)

Comment 12

[Eric Pollmann]

Percent height fixes checked in.  Chris, this was the one-liner you suggested 
when you were up here and I had Harish look at it.  I also checked in a fix for 
CalcQuirksModeHeight for native scrollbars, and added one sanity check.  Will 
attach testcase shortly.

Comment 13

[Eric Pollmann]

Attachment: reduced test case

Comment 14

[Chris Petersen]

Fixed in the July 6th build.