AltME groups: search

build#53 committed to SVN. Button widget rewritten to take full advantage 
of AGG and fix some anomalies (e.g. actions now fire on button up 
instead of button down, replaced 'active facet with options [info], 
hover effect no longer "sticks" when a popup is called from a button, 
etc). And all this in 1/4 of the code! ;)

Robert, is there a tree widget available yet in your version?

I thought there was a tree widget embedded inside another widget 
of Cyphres

The spinner widget now using alt click .. is that a problem for OSX 
?

This seems to be slightly anomalous behaviour.  When a modal requester 
is present, every other widget works ( in other windows ) except 
the drop/edit list.

spinner widget now using alt click .. is that a problem for OSX
 Nope, OSX maps that to CTRL-click

The build I'm working on reverts spinner to a more "normal" behavior 
(I've worked out how to get half height arrows)
slightly anomalous behaviour ... modal requester

 The underlying View popup/modal system has some subtle bugs, some 
 of which will hopefully be fixed in R3.
scattered 

attempts" throughout the Rebgui code" ... Odd, I've never needed 
this. Is it only in more recent build(s)? I'd really like to identify 
and fix the underlying cause(s).

Hmm.  The new version of process-keystroke has a lot more code in 
it.

Causes? So far you've mentioned three:

	Clicking too fast/repeatedly in a field
	Sporadic area errors
	Tab-panel with focus in multiple panels

All these point back to the edit object.

in 'edit-text .. I have a large attempt block from		caret: caret-to-offset 
face view*/caret to show-face

ie. the chevron was a date change by day, and the arrow was by week.

build#55 committed to SVN. Numerous widget fixes and enhancements 
to the following widgets:

	anim
	
arrow
	
bar
	
box

	button
	
field
	
group-box
	
image
	
label
	
pane
	l
progress
	
spinner
	
splitter
	
text
	
tab-pane
	title-group


The most noticeable improvements include the following tab-panel 
enhancements:

	first tab action now fires when widget is first displayed

 can select an alternate tab to initially display with "options [tab 
 2]" etc

 can programmatically select a tab at runtime with var/select-tab 
 2

title-group and panel now have rounded corners.

Ashley, yes, but we first have to merge to a stable release. I'll 
talk to Cyphre how to do it. Because we did it on our side but some 
changes we made are not compatible. So let's see where we are and 
than decide how to merge it.

Graham, the code is there but not yet put into a seperate widget. 
But I think it could be done very easy.

Works for me. Try putting an "attempt [delete %ui.dat]" prior to 
the "do %rebgui.r" line to see whether it's a config difference.

re: chevron. You'll also be glad to know that chevron will re-emerge 
as an option of arrow. The previous implementation didn't take advantage 
of AGG's new line-width command,and duplicated much of the code found 
in arrow anyway. I'm putting RebGUI on a diet and trying to remove 
some of the "fat" that has managed to creep in. ;)

Haven't figured out why tabbing is screwed though in my app. so going 
to use action blocks to direct tabbing as a work round.

I think it would be good to allow a spinner to activate the action 
block

And to reproduce the error, I just clicked in the first field here 
..

do %rebgui.r

display "" [

 label "field has a focus action" a: field [ show-focus b ] return
	label "also has action" b: field [ print face/text ] return
	label "no action" c: field return
	spinner options [ 9:00 24:00 0:30] [ print face/text ] return
	button "Quit" [ unview/all halt ]
		
]

do-events

I've reproduced the tabbing issue .. but that appears to be from 
some of my own changes as it is not present in a latest checkout, 
but is present in the checkout I did manually.

Ok, found it.  If action-on-tab: [ field ], and the field action 
is show-focus as in the example above, then it tabs from a => c

Don't know why all my code that has {show [widget]} now fails with 
#55... but I can work round it by  ... wrapping a condition around 
it

if object? widget [ show [ widget ] ]

different topic .. is it better to hide a button, or, to gray it 
out ?

if the latter, should a button have a grayed out effect where the 
actions are disabled ?

display "test" [
	button options [info]
]

to display a disabled button. Can't set that dynamically though.


Couldn't reproduce the editor error above, but the tabbing "error" 
is a sequence issue (it executes the show-focus then the tab).

what I saying is that before #55, all my code worked.  With #55, 
if I have code that does a show on a widget, it may fail

Is copybtn a hidden button that your action tries to make visible?

Can you reduce the problem to a small sample and post here? Layout 
hasn't been changed but the way in which tab-panel initializes has.

On another topic, I often find I need symbols such as those in a 
music player (stop, forward, fast-forward, etc) and in lieu of a 
cross platform "symbol" font I'm thinking that a symbol widget using 
AGG might be the go. Would work something like:

	display "test" [
		symbol data 'square
		symbol data 'arrow-right
		symbol data 'double-arrow-right
		symbol data 'circle
		symbol data 'circle options [no-fill]
	]


so the two questions are; would this be useful, and if so what would 
constitute a good set of symbols and names?

what's wrong with this?
 display expects a unique title string. ;)

Try this one ...

rebol []

do %rebgui.r

show-bug: does [
	display "test window" compose/deep [
	tab-panel 80x20 data [
		action [ hide e ]
		"1" [
		]
		action [ show e ]
		"2" [
		]
	]
	return
	e: button "Invisible"
	]
]


display "Test"  [

 label "field has a focus action" a: field [ print face/text ] return
	label "also has action" b: field [ print face/text ] return
	label "no action" c: field return
	spinner options [ 9:00 24:00 0:30] [ print face/text ] return
	button "Display tab-panel" 40 [ show-bug ]
	return
	
	button "Quit" [ unview/all halt ]
		
]

do-events

Yes. Wait a moment and I'll get another build with the tab-panel 
and spinner fixes up ...

I've just noticed something odd with tabbing inside a tab-panel. 
It tabs through all widgets on every pane!? I need to look at Cyphre's 
edit (cyclic?) changes I think.

keyboard navigation is however the only way to drive a gui at full 
speed .. by ignoring the gui!

can the increment value in a spinner widget be changed dynamically?

but of course there is a lot of training required to learn all the 
keys ( shift, control, alt on every key )

all done by keyboard navigation.

 as long as the widgets don't require much internal navigation. There's 
 a world of difference between a screen with 12 fields and one with 
 drop-downs, tables, scrollable areas, check boxes and radio buttons 
 (to name a few).

I suggest that for spinners, allow users to start with a blank spinner 
so that they can type in their own initial values.

Agreed, the assumption is that data must start with a valid default. 
I'm looking at how to enable data to default to none and text to 
"" ...

Group-boxes, for me, have a visual anomaly now where the left edge 
eats into the label text of group-box

I've re-encapped all my apps with the latest build and they all idle 
at 0-1%. %tour.exe idles at 50-75% though! Hmm, I might cut widgets 
out one at a time to see if it's a particular widget (e.g. anim or 
a timer) or just the number of widgets/events that's the problem. 
Good catch (I've been more concerned about optimizing for memory 
use of late, not CPU).

typing felt a little sluggish and I was exploring why!

Graham, Pekr probably mean, that it's nice that you announce build 
numbers. But for example I don't know how to access your svn so the 
build numbers are a little bit useless. So maybe one day it will 
be fine to see some link into rebol file as well:)

Yes, adding some RebGUI stable release to a web-site is a good idea. 
I can do it, or Ashley do you want to do it on your web-site?

well, I downloaded SVN (Tortoise) to not be a lazy dog :-) I just 
don't know what to do with it yet :-)

trac is a different program altogether .. it's a web interface to 
svn

right click on a directory, and then choose checkout from the svn 
directory

Oldes - you have a better design we can use?

before there was a thin highlight to show which tab you were on ; 
now the whole tab is highlighted.

Ok, need to create a english.dat file and place into the dictionary 
folder.

Ashley, how about including a dictionary directory, with english.dat 
in the svn?

will you guys build downloadable file once you reach merged and kind 
of stable release?

 Last stable build is #46 available via the View desktop and here: 
 http://www.dobeash.com/download.html


Once the current changes "settle down" (i.e. at least a week or so 
passes without a major new issue/problem) I'll update the stable 
set.

161kB.. it's pretty large
 ... 30+ widgets tends to do that! ;)

I like the BTN style of button which is in Rebol by default

 ... Come up with an AGG (not image-based) equivalent or similar and 
 I'll gladly use that.


re: tabs (and button) look. My previous goal was to try and map the 
look as closely as possible to WinXP. This required a combination 
of images and complex draw commands. My goal now is to keep it as 
simple as possible, with a nice clean look that can be implemented 
with as few effects/draw commands as possible. Button is an example 
of that. Instead of using 3 images and changing them based on current 
button state, I now use a simple draw block and change a single value 
based on state. Note that the radius is customizable (via effects/radius). 
Does this produce the best looking button ever? No. But don't fault 
the technique, rather my [limited] AGG compositional skills! Feel 
free to come up with a better button algorithm.


tabs are another example, where yellow lines of varying length were 
drawn to approximate the WinXP tab look, and had to be cleared and 
redrawn based on state changes. The new approach uses a simple effect 
block ( [round color 5] )where all that needs changing on a state 
change is the color. Same deal as button applies. Come up with a 
simple draw block that creates a good looking tab and I'll gladly 
use that.

Few minutes ago I talked with cyphre, and he told me, there is a 
tree-view widget for rebgui, as he separated it from other widget 
code. IIRC Graham asked about tree-view widget few days ago. So, 
if Robert will agree with release, cyphre will upload it ...

Did the default behaviour of radio-groups change recently?  I think 
if you did not specify a number, or none in the data block it used 
not to set the first radio.

It changed slightly with Cyphre's changes in that not specifying 
a number (or none) now defaults to selecting the first item. If you 
want to start with no items selected then you must use none (0 does 
the same thing, but causes problems for picked and selected until 
a "valid" item is selected ... I'll fix that for the next build).

Has drop-list changed?

 That got a big work-over with Cyphre's changes. Many problems were 
 fixed and enhancements made, unfortunately it seems to still have 
 problems in some specific circumstances relating to where the drop-list 
 is located on the display and how many items are to be displayed. 
 I've "cleaned" (i.e. examined every line of code in detail for errors 
 or possible optimizations) a dozen or so of the simpler widgets in 
 the previous build(s), drop-list is a high priority for cleaning 
 in the next.

radio-groups: Ok, maybe my english is bad. Let's try with an example:
	radio-group ["Option 1" "Option 2"]


Now I have to save the user selection into a database. So I store 
"Option 1". 

While further developing my app the code becomes:
	radio-group ["Option A" "Option B" "Option C"]

Now:

- How to restore an older data-record with"Option 1" which now becoma 
"Option B"?


If we use IDs I can decouple the texts from the stored values in 
the DB.

since radio-group/selected saves the associated text, that might 
be a better way to store the choice.  Then make a complementary radio-group/select-item 
n [ integer! string!]

Seems like a binding problem.

The difference is, you have level as a local to your function, I 
have level as a member of the object. What do you give to build-markup? 
It's a string, right? So build-markup don't know, what you mean by 
level, I think.

Yes, I give it a string. But changing to "return build-markup {<%html-handler/level%>}" 
raises the same error.

Look at it from build-markup's viewpoint. How should it know, what 
level mean? This is a bit tricky part of REBOL, but it's also one 
of REBOL forces.

In my last example, the string become something like {<%3%>}

In your example, the string is {<%level%>}. This is what build-markup 
see, and it has to figure out, what level mean. It can't figure that 
out, because level is a local of your function.

All functions are in a situation like this. {<%level%>} is just a 
string giving to a function.

I've never seen such a behavior in whatever langage I've used, but 
I'm starting to understand it : thanks for the clarification.

Maybe you can think of it as two persons, one giving the other a 
string. The person receiving the string read "<%level%>". That person 
now have to figure out, what level mean. The person live in a context 
(the global context in this situation, because everybody can call 
build-markup). So it look up level in the global context and find 
nothing, so it gives an error.

If you instead write: rejoin ["<%" level "%>"]

then level is replaced by the value of level and a string is made, 
and this new string is sent to build-markup.

Rebolek: I meant "was it a good alternative to VID ?". The only thing 
I miss in every Rebol GUI I've seen is font antialiasing.

font-A: make face/font [size: 32]

view layout [box 200x100 black effect [draw [font font-A text "Hello 
World!"]]]

view layout [box 200x100 black effect [draw [pen red fill-pen yellow 
font font-A text "Hello W
orld!" vectorial]]]

right, you figure that part out and make a nice GUI for us! :-)

No problem! give me just a min ... or a year :)

No, NodeBox is for poets : http://nodebox.net/code/index.php/Gallery
. Have a look at the Helsinki gallery : it's really beautiful.

I don't know, if anyone has done something similar in REBOL. There 
are demos flying around. I made a very simple Bezier demo once:

do http://www.fys.ku.dk/~niclasen/rebol/fysik/bezier.r

It could be a start to make something better and more beautiful.

hehe, they may be from a contest a few years ago. You can find them 
on the ViewTop under REBOL.com/Demos/
RZoom is also nice (by Henrik).

I'm amazed by cyphre source file : just a few lines.

I've made a C DLL on linux which has the rudiments of an image processing 
function. At the moment it just sets one pixel.

I never had a problem with it since a month. It's really stable.

Anton, have you seen new speedy version of my tunnel script ? :)

Lookup table for pixels position and tunnel coordinates helped a 
lot...

I guess your dll would help. Would you keep acceleration only for 
pixel drawing? or maybe some higher functions?

Oh no! Calling a DLL function just to draw one pixel would be so 
slow! I will of course make it process a whole frame at a time. So 
basically, I think the tunnel drawing function will move into C.

Yes, in that one I used AGG to scale and composite a relatively low 
number of frames, zoomed to different distances.

Pekr, on my old machine, I was hitting the limits of detail I could 
put in without reducing the framerate too much. I had to tweak it 
a lot to keep it looking fast and detailed enough. It was difficult.

Yes, Lotus Turbo Esprit I played a lot.

Tiny C Compiler (TCC).

I've installed an older version of TCC and tried compiling a simple 
example program, but it failed to compile, and the error messages 
led back to the above-mentioned problems, where patches to fix the 
compiler can be found in a non-main maintainer's forked repository, 
which is no longer supported by him.

Hi. Is there a way to call a method (an object function) programmaticaly 
? ie supposed I've got a Car object with a function "drive-to" inside. 
Now, I've got a string "action" equal  to "drive-to". I want to do 
Car/action but that does not work (I can understand why), but is 
there any workaround ?

(I never used such a code before:)

IN is a nice way to tell whether a specific word exists in an object:

if in word 'foo ['bar]

hi i would like to give you a example i am trying to do of a CRUD 
application

that create a folder and you must  execute the script with rebol 
START.R

it is just a simple example of a create refresh update delete of 
a pretty db rebdb with rebgui

with entity and "percistence"  i would like to have a java EE5  but 
in rebol or better in R3

Hi. Is there a build-in function do do something like this : given 
the number 5 and the string "abc"  construct the string "abcabcabcabcabc" 
(equal 5 times "abc")?

And if you're really concerned about speed, this is a little faster:

to string! array/initial 5 "abc"

Well, array just give a block. Maybe you didn't want a string but 
an issue:
>> to issue! array/initial 5 "abc"
== #abcabcabcabcabc

or something else. Strings are just one form of series.

Have you got a look at the XeLaTeX output ?

A motto I like very much and think about daily:
Everything should be made as simple as possible, but not simpler.
- A. Einstein

And then just wrap it up:

    string-of: func [
        "Returns a string of <value> repeated <count> times."
        count [integer!]
        value [word! char! any-string!]
    ][
        head insert/dup copy "" value count
    ]

BrianH and Gregg : thanks, I need to write all these tips on a page 
now.

Geomol: more than ever. But I'm rather fed up now working on this 
markup engine. I need fresh air.

I was thinking of a game, and in fact I've found one, but I think 
I need your help. Beware : that's not easy !

:-) We have a game group. You could start posting there and see, 
what will happen.

How could I generate a random number (a real number) between 2 given 
values, ie between -pi and pi ?