[API Review] : Baseline offset depends on layout bounds which are calculated during layout

Martin Sladecek martin.sladecek at oracle.com
Mon Sep 9 11:11:23 PDT 2013

Hi Scott,
I don't know why it was decided that non-textual Nodes have baseline == 
height, but I suppose it was because HTML does the same.

<table border="1">
    <td valign="baseline">abcde</td>
    <td valign="baseline"><div 
style="width:100px;height:100px;border:1px solid;"/></td>
    <td  valign="baseline" style="font-size:200%">fgh</td>

Will align the text baseline to the height of the "div" rectangle.
Maybe it's possible to find some explanation for HTML.

Note that the descenders are not cut off, as it's just the height of the 
Node, not the full height of the HBox/Gridpane row/etc... The full 
height will be the Node's height + descent.


On 09/09/2013 06:00 PM, Scott Palmer wrote:
> I don't get it. When is the height ever the correct position for the 
> baseline?  The height value can't actually be used as the baseline as 
> that would mean descenders are cut off.
> How can the layout container compute the baseline for a node if it 
> isn't what is painting the text?
> Scott
> On Mon, Sep 9, 2013 at 10:19 AM, Richard Bair <richard.bair at oracle.com 
> <mailto:richard.bair at oracle.com>> wrote:
>     Sorry we flew off the list accidentally. I like
>     BASELINE_OFFSET_SAME_AS_HEIGHT. This change in behavior will
>     potentially break other layout managers which are handling the
>     baseline case. Specifically, we should test with MigLayout.
>     Otherwise this seems like a significant improvement in that we can
>     avoid multiple passes and also correct layout the first time
>     around instead of having things in the wrong place the first time,
>     only to be moved around the next pass.
>     Richard
>     On Sep 3, 2013, at 9:50 PM, Martin Sladecek
>     <martin.sladecek at oracle.com <mailto:martin.sladecek at oracle.com>>
>     wrote:
>     > Maybe the USE_COMPUTED_BASELINE_OFFSET should be rather
>     > It's up to the layout container to compute the baseline of such
>     Node, as the layout container must compute the height as well.
>     > Currently, the layoutBounds height is returned, but this is
>     incorrect, as the layoutBounds contain the old width&height and
>     will be updated during the layout.
>     >
>     > So it actually means, for the baseline of this Node, use it's
>     newly computed height.
>     >
>     > -Martin
>     >
>     > On 09/03/2013 09:45 PM, Richard Bair wrote:
>     >> I must be missing something. I thought the getBaselineOffset
>     method was the means by which that was done? Meaning, if I'm a
>     layout container, then I would call getBaselineOffset to get this
>     value. Now, I call it and it tells me
>     USE_COMPUTED_BASELINE_OFFSET, so how do I compute it? Or do you
>     mean, if somebody gets USE_COMPUTED_BASELINE_OFFSET, then it
>     really means, use the layoutBounds height?
>     >>
>     >> Richard
>     >>
>     >> On Sep 3, 2013, at 12:37 PM, Martin Sladecek
>     <martin.sladecek at oracle.com <mailto:martin.sladecek at oracle.com>>
>     wrote:
>     >>
>     >>> The layout container will need to compute the baseline for
>     Nodes with USE_COMPUTED_BASELINE_OFFSET, because it's the layout
>     container who needs to compute the child's height (which is equal
>     to it's baseline). I've prepared the code of all the layout Panes
>     so that the baseline is computed after the heights and the heights
>     are computed with respect to the baseline. There are many
>     package-private helper methods in Region that I've updated and
>     used in our layout Panes, so it's a matter of a few calls for us,
>     but 3rd party layouts will have to compute this by themselves (as
>     they aways had).
>     >>>
>     >>> -Martin
>     >>>
>     >>> On 09/03/2013 05:56 PM, Richard Bair wrote:
>     >>>> So how does the layout container ask for the computed
>     baseline offset? It used to be, by calling this method.
>     >>>>
>     >>>> Richard
>     >>>>
>     >>>> On Sep 3, 2013, at 6:11 AM, Martin Sladecek
>     <martin.sladecek at oracle.com <mailto:martin.sladecek at oracle.com>>
>     wrote:
>     >>>>
>     >>>>> Hi,
>     >>>>> related JIRA issue:
>     https://javafx-jira.kenai.com/browse/RT-31006
>     >>>>>
>     >>>>> I propose to add constant "public static final double
>     >>>>>
>     >>>>> This would be returned by the default Node implementation of
>     getBaselineOffset. Currently, the layout bounds are returned, but
>     they are not computed at the time getBaselineOffset is called,
>     which leads to incorrect layout and other problems (see more in
>     description of the issue).
>     >>>>>
>     >>>>> Thanks,
>     >>>>> -Martin
>     >

More information about the openjfx-dev mailing list