World: r3wp

It only has a title extra over a non-crashing version

I get this from your version of the demo:

>> do %r3-gui.r3

Script: "R3 GUI - load and start" Version: $Id: $ Date: 9-Dec-2010/10:32:04+1:00
>> do %demoJC.r3

Script: "R3 GUI - Development Test Script" Version: 0.1.1 Date: none

Script: "R3 GUI - load and start" Version: $Id: $ Date: 9-Dec-2010/10:32:04+1:00
** Internal error: stack overflow

** Where: reduce switch parse to-text reduce parse to-draw all update-subgobs 
foreach update-subgobs foreach update-subgobs foreach update-subgobs 
show-native show-native show-native show-native show-native show-native 
show-native show-native show-native show-native show-native show-native 
show-native show-native show-native show-native show-native show-native 
show-native show-native show-native show-native show-native show-native 
show-native show-native show-native show-native show-native show-native 
show-native show-native show-native show-native show-native show-native 
show-native show-native sho...

At one point there was a guard added to make sure that SHOW-NATIVE 
was only redefined once, to avoid that stack overflow.

If I load RMA's original GUI in Jocko's R3, I get the same result

I also have that. If you launch directly %demoJC.r3, you don't have 
the problem (there is a call to r3-gui.r3 inside demoJC.r3). It seems 
that if you call it two times, it crashes.

to summarize: open a new cosole, launch directly demo:
do %demoJC.r3, and it should work ok.

OK, that works

But it also segfaults when moved

not mine ... strange !

On WINE. Also when just using the demo

Unfortunately, I can't show this at ReBorCon

This GUI and GLASS are also terribly slow

I have to leave, now
Anyway, I wish you a fruitful conference.
bye

Thanks for the help

Mine is here - http://xidys.com/pekr/rebol/demo.r3

not much time to fix everything anyway. There's two fixes for Doc 
style but generally - this gui is incompatible to Carl's one in many 
aspects ....

Does not work with A111. Carl should really not do rushed releases, 
if he has no time to push R3 further forward!

I have to say, I'm surprised that you endured this for weeks

well, not sure I understand the word "endured" correctly, but anyway 
- I wanted to become more familiar with the GUI, published some tickets, 
learned a lot, yet I am far from being a good coder. I also only 
have few hours a week to experiment. I initially thought it is going 
to be about changine few things here or there, but it seems to be 
more complicated.

I believe the released GUI is out of sync with A111. The GUI works 
here.

I think we're talking different definitions of "works" here

One or two years ago, Carl's GUI was demonstrable. This one isn't

Well, hopefully the next release will be more stable.

Pekr, we are by and large at the same point. Your version is compatible 
with the a111 version that I compiled, except for a certain number 
of tests, like mine. Did your version work with the a110 of RMA for 
the drawing test ?
Kaj, I agree, Carl's version was demonstrable.

Some basic functions like request or alert, and styles like doc are 
still bugged in r3-gui.r3

Some basic functions like request or alert

 - because they are an entirely separate scheme that is not yet implemented. 
 Carl's version of these were unorganized and lacking.

jocko - you can fix Doc style - look at following ticket  - http://curecode.org/rebol3/ticket.rsp?id=1850&cursor=10

pekr - thanks, I will do it when I have time.

I have put a new version of test-demo, where the status of what is 
working and what is not is mentioned, and the crashing actions and 
faces are disabled. Pekr, I have not yet integrated your fix for 
the doc style ... later
http://www.colineau.fr/rebol/downloads/tests-R3Gui.zip

pekr - I see that you have put many other contributions on r3-gui 
bugs

well, some of tickets are not clearly a bugs, but misconceptions 
... which most probably might get dismissed because of my incorrect 
understanding of the topic. But I wanted to have it "recorded". I 
just don't know, if RMA works with those tickets, as e.g. Doc style 
fix is not yet commented. I will wait for new release and we will 
see. Hopefully together we might get the demo working. I expect the 
biggest problem being in the area where new box model is involved 
... we will see ...

I agree with you. The point is, as there is not enough documentation, 
we may "translate" incorrectly the code. Anyway, I would be pleased 
to contribute with you to make this demo work, as I feel it very 
"sexy" and representative of the potential of the gui. Of course, 
RMA is welcome to do on their side this kind of demo, showing the 
various functionalities of r3-gui

Guys, I don't know how often I have to repeat it:


1. we get basic concepts implemented. All these basics are seperated 
as much as possible. That's why it will be totally incompatible to 
Carl's GUI. And, we don't try to be compatible. Our goal is to have 
an enterprise enabled GUI framework.


2. To do step 1, we just pick 1-2 styles, and enhance and change 
them exactly to just work with the concepts from 1. And, please see 
the plural in "concepts". We are doing step 1 & 2 by concept.


3. If this stabelized, we take the next style and adopt it. This 
might lead to some changes in 1. which than have to be carried forward 
again.

4. Then either next style or next concept.

It's quite simple. The GUI is not functioning in general purpose. 
Anyone expecting this is just pure wrong... you have to wait or help. 
And the concepts and styles we care are priorized by what we need.

What we're observing is a regression over years

I wasn't expecting functionality, but I was expecting demonstrability

Well, RMA doesn't work for years on R3-GUI.

It's been a year now

The other R3-GUI stuff is nothing I take seriously into account for 
now.

Yes, but not full-time. As we first did some tests, to proof if we 
can invest into the R3 line.

It's one of the base rules of software engineering to keep complex 
systems in a working state

The reset of R3 development was already a flagrant violation, the 
results of which we're dealing with now

It works for us for those things we focus on. Of course, if you try 
to do other things it won't most likely not.

I agree, that's why we bite the bullet and invest a lot of money 
into the development of a useable GUI framework. It's a hell lot 
of work as we mostly started from scratch, and need to fight on the 
c-level as well.

But it doesn't work. And you'll have great trouble getting back to 
a working state, with so many issues interfering. This alone explains 
the long time spans

Well, you seems to know more than me than. Anyway, yes, than that's 
the case.

I would hope not

Maybe it would have been better to have kept Carl's GUI as something 
that could have been used warts and all, .. and the GUI team continue 
to work on their alternate GUI?  Would that have been possible?

That means more work, so it would have to be handled by people other 
than Carl or RM Asset.

Carl's GUI is not working because of internal changes like moving 
graphics to extensions module (using commands)