Further, the current SystemWindow>>addMorph:fullFrame: assumes thick borders between components, for pane splitters.  There used to be an accessor on SystemWindow, allowPaneSplitters:, that could be used to discard pane splitters and their thick borders, but this was ditched.

So if I want a window containing lots of morphs with no splitters and no thick borders how do I do it?

a) Should I use a container morph and add this to the SystemWindow?

b) should I add an override to SystemWindow that does what addMorph:fullFrame: (or addMorph:frame:) does without adding the thick borders?

c) should I add back the allowPaneSplitters[:] accessor and modify SystemWindow>>addMorph:fullFrame: to take account of allowPaneSplitters == false?

a) seems the right approach, but when I use either a Morph or a BorderedMorph as my container I don't get things laid out correctly in that morph.  Is there a standard container morph that does layout correctly?  If so, what is it?

c) is easy to do, but folks have to maintain it (it's no more than a guard around all the frae manipulation in addMorph:fullFrame:).

On Wed, Jun 12, 2013 at 3:23 PM, Eliot Miranda <eliot.miranda@gmail.com> wrote:

On Wed, Jun 12, 2013 at 2:39 PM, Eliot Miranda <eliot.miranda@gmail.com> wrote:

Hi All,

    in SystemWindow>>addMorph:fullFrame: towards the bottom are the lines:

aMorph adoptPaneColor: self paneColor.

aMorph borderWidth: 1; borderColor: Color lightGray; color: Color white.

I'm adding a UpdatingThreePhaseButtonMorph checkBox to my window so, via ImageMorph>>colors: 

ImageMorph>>color: aColor

        super color: aColor.

        (image depth = 1 and: [aColor isColor]) ifTrue: [

                image colors: {Color transparent. aColor}.

                self changed]

this has the effect of completely smashing the checkBox's normal image, and setting to white ScriptingSystem formAtKey: #CheckBoxOn, which is what the checkBox uses as its default image.

What business does SystemWindow have of smashing the colour of any morph added to it?  Surely this is bogus, no?

I see the regression.  Back in 3.9 the method read

(aMorph isKindOf: ImageMorph) ifFalse:[

aMorph adoptPaneColor: self paneColor.

aMorph borderWidth: 2; borderColor: #inset; color: Color transparent]

Unless anyone objects I'll put back this guard.  it seems obviously correct (or rather, obviously less broken; the method also includes this priceless gem:

(aMorph class name = #BrowserCommentTextMorph) ifTrue:

[aLayoutFrame rightOffset: windowBorderWidth negated.

aLayoutFrame leftOffset: windowBorderWidth.

aLayoutFrame bottomOffset: windowBorderWidth negated.

aLayoutFrame topOffset: (windowBorderWidth negated) + 4].

).

--
best,

Eliot

--
best,

Eliot

--
best,

Eliot