World: r3wp

Yeah, all these points do need to be addressed.  I will hold him 
[Mr. C] down, you tickle him.

View's modal system can handle closing on moving the mouse away, 
you just need to set a custom handler.

I think requiring a click outside is good behaviour, but it is only 
good if that click is not consumed, but passed on (so that the second 
drop-list  activates straight away)..

Agreed.  


One more subtle point, on Windows, both in Firefox browser, and in 
products like Word, the behavior is:


Moving your mouse over Droplist-1 highlights it (this is true of 
all widgets)
Droplist-1 is clicked on, and opens.

Moving the mouse to Droplist-2 does not highlight, because the first 
widget is engaged.

Clicking on Droplist-2 closed Droplist-1, and highlights Droplist-2, 
but does not engage it (open Droplist-2).


Of note, even something as simple as a checkbox on Windows has 8 
states:

Unchecked
Checked
Unchecked highlighted
Checked highlighted
Unchecked ghosted 
Checked ghosted
Button down unchecked
Button down checked


I strongly vote copying the most used UI in the world on every little 
detail.  They might not be "right" but they are the popular vote.

RebGui has some rough edges, but over all it has been really easy 
to build stuff with.  Ashley, on our Qfile client application, we 
are putting in the Credits RebGUI, and a ink to your site.  We will 
post a beta on Qtask and if you want something changed just tell 
us.

It would be good if changes can be made to the svn store.

What do you mean exactly, and what interface to SVN do you use? (I'm 
sort of studying iCVS vs SVN right now).

one uses a svn client.

I believe I posted the details on qtask.

And if I didn't, Ashley has a list of clients on his main page at 
http://trac.geekisp.com/rebgui

Do you use TortoiseSVN?

I started out with Tortoise SVN, but it doesn't work well with DirOpus 
for windows.

So, I'll be looking for a non GUi client.

Interesting.  I made a joke to a freind of mine about how I keep 
getting depressed learning about the features of some given product 
that is version 2.  I look at the product that is version 2 that 
is missing important features, and then wonder, what did version 
one have?

Version 1, our light switch features "on." 

Version 2, in addition to the much loved "on" feature, you can now 
turn the light "off"

I really feel this way about SVN and CVS.

Might be related to the discussion above:

If I have a 'field that uses an on-unfocus event and I enter some 
test and than click on a drop-down list, the on-unfocus event for 
the field isn't generated.

Cyphre: The download is very slow... just for your information.

How can I create something like a STYLE to avoid always having to 
specify things like
	font [align: 'right]

view layout [style my-field field font [align: 'right]  my-field 
my-field]

or you could restyle the field with 
	layout [style field field font [align: 'right]  field field]

Does RebGUI support the 'style  keyword? Ah, didn't knew it.

Whoops! Sorry, didn't pay attention to the group name... :-(

But I suppose you could do something like:	
	rebgui-widgets: make rebgui-widgets [

  my-widget: make one-of-the-widgets [font: make font [align: 'right]]
	]

What exactly was Cyphre's change?

***** rebgui-widgets.r *****
Lines which are not present in the other.
-------------------------------------
419 :       show-popup/window popup parent/parent-face
420 : ;     show-popup/window/away popup parent/parent-face
***** rebgui-widgets.r *****
Lines which are not present in the other.
-------------------------------------
407 :           feel: system/words/face/feel
420 : ;     show-popup/window popup parent/parent-face
421 :       show-popup/window/away popup parent/parent-face
618 :                   down    [face/data: on]
619 :                   up      [face/action face face/data: off]

thanks.

I hate this line.  I hate this line.  I hate this line:

		either 1 = length? view*/screen-face/pane [quit] [

I hate this line.
By the way, I hate this line.

Here's a patch to group-box which makes the top line font-size sensitive:

;init:	make function! [] [ ; <- this multi-arg spec notation will 
be broken in rebol 3
		init:	func [] [ ; <- but this will still work
			font/color: colors/menu

   ; note: if you recalculate the draw block in feel/redraw instead, 
   that allows dynamic font-size changes
			insert effect/draw compose [
				pen (colors/edge)

    ;box (as-pair 0 sizes/cell * 2 - 2) (size - 1x1) ; -2 accounts for 
    para/margin/y, I think
				box (as-pair 0 ((second size-text self) / 2)) (size - 1x1)

    ; blank out the portion of the top edge of the box where the text 
    is
				pen none ; <- this will be set later by feel/redraw

    ;line (-2x-2 + as-pair sizes/cell * 2 sizes/cell * 2) (as-pair sizes/cell 
    * 2 + first size-text self sizes/cell * 2 - 2)
				line 

     (as-pair (sizes/cell * 2 - 2)					 ((second size-text self) / 2)) 
     ; -2 for space between line and title text (on the left side)

     (as-pair (sizes/cell * 2 + first size-text self) ((second size-text 
     self) / 2))
			]

-2 accounts for para/margin/y, I think
  <- you can ignore this :)

have you thought of asking Ashley for a write access account to the 
svn?

Anton, you could try some buddhist lotus meditation, and suddently 
you'll hear wonderfull music, smell exotic odours, see never-thought-of 
colours ... and that line will look very beautiful to you! ;-)

Graham, no I haven't. I don't think I'm so closely involved.

Geomol, I was enlightened to hit the public cache with a few ;;;;;

That line you hate:

	either 1 = length? view*/screen-face/pane [quit] [

could be replaced with:

	unless 1 = length? view*/screen-face/pane [

better?

That just basically removes the quit from the equation ? Should be 
ok. :-)

tables: Is it possible to change the column layout of a table after 
it was layouted? I have a space where I display a table that changes 
it's layout depending on the context of the app. Sometimes it has 
2 than 5 columns and different column labels.

tab-panels: If I use nested tab-panels and add a pre-enter action 
for the nested tab, it's not executed for the first tab shown. As 
this tab is implicitly shown if the parent tab is clicked, the action 
doesn't seemed to be triggered.

Ashely didn't agree that this was okay.  I suggested it before.

Not quite, it was logged as issue#21 ... its just never been high 
on my priority list to fix.

I thought "under consideration" was an euphemism for "don't agree" 
:)

Ok, going to look into it together with Cyphre.

If you're inside an area field with some text, and a slider showing, 
and if you position the slider to the bottom, and then click inside 
the area, the text scrolls back so you're not at the bottom anymore.

Probably the caret hasn't been updated to reflect the position where 
the scroller is.

.

Sorry about that... I just got dropped. :-(

Ok, submitted a ticket to the trac.

There's some funny business going on with key-scroll?. If move the 
caret using arrow keys sometimes the next click on the scrollbar 
doesn't work.

Graham, this patch to the area/pane (slider) action makes things 
better:

				action:	make function! [face /local pos new] [

     ;	don't alter scroll if the last function to touch it was edit-text!!!
					unless parent-face/key-scroll? [

      parent-face/para/scroll/y: negate parent-face/text-y - parent-face/size/y 
      + 3 * data

						if system/view/caret [
							system/view/caret: 
								offset-to-caret parent-face 
									new: min max 
									(pos: caret-to-offset parent-face system/view/caret)

         0x0 ; <-- should add height of one line of text, to keep caret fully 
         visible

         parent-face/size - (face/size * 1x0) ; <- should subtract height 
         of one line of text
						

       ;print [parent-face/size pos new parent-face/para/scroll/y index? 
       system/view/caret]
						]

						show parent-face
					]
					parent-face/key-scroll?: false
				]


It also needs this patch to work correctly, near the top of rebgui-widgets.r:


 ; Unfortunately, offset-to-caret returns end of the string when offset 
 is between two lines, 

 ; which is only possible when indent/y > 0. This ought to be submitted 
 to rambo as a rebol/view bug.
	; I would not use indent until it is fixed.

 ; offset-to-caret needs to work correctly to allow the new area widget 
 functionality of keeping
	; the caret visible when scrolling. -Anton


 ;default-para-wrap: make default-para [origin: 2x0 indent: 0x2 wrap?: 
 true]

 default-para-wrap: make default-para [origin: 2x0 indent: 0x0 wrap?: 
 true]

Key-scroll? issue I mentioned above there is apparently fixed by 
adding this line to area/feel/redraw:


     ;	don't alter data if the last function to touch it was slider!!!
					if all [face/pane/ratio < 1 face/key-scroll?] [

      var: either view*/caret [-3 + second caret-to-offset face view*/caret] 
      [0]

      face/pane/data: either zero? face/para/scroll/y [0] [(var + negate 
      face/para/scroll/y) / (face/text-y - sizes/font-height - 2)]
						face/key-scroll?: false ; <-- added this line -Anton
					]

Anton, did you made the changes to the TRAC repository as well?