World: r3wp

Gabriele - sorry, but it seems to me, that your pov is misleading, 
because you try to base it on incorrect presumption. There were some 
rules set. The rules were - Carl decides, I coordinate, write down. 
Sorry, but if someone disappears for 2 - 3 months, then morale of 
group can't be kept high. I am not blaming Carl for anything here, 
that was not my point, but I have to point to the reason of failure 
...

in fact, "design by steering commitee" is kind of broken record from 
RT's side. It would be better to admit, that someone is not used 
to work in real team. That is still fact, but it should be admited, 
or false presumptions are accepted then ...

the truth is, that it si you, who started to talk about commiteee. 
And it is not imo fair to this group, because it was RT, who first 
outlined VID 2 group, as closed group of VERY FEW developers. And 
some of us, remembering Henrik, Anton, expressed their will to accept 
such group invitation ...

so, do you call 3 - 5 ppl group to be a commitee? Besides that, the 
intention of the group was planned just to outline what new VID should 
be, actual implementation could be done by even less ppl.

What I actually see is you collecting ideas, e.g. from Maxim, discussing 
some things, but admitting you will choose only some parts of eventual 
data-flow aproach, maybe without understanding whole Maxim's engine 
purpose? I don't really mind new VID being your or Carl's only decision, 
but what I would regard as being fair is - create some document about 
planned architecture, and let it run via 1 or 2 round of comments 
here. Actually, you prepared your original ideas in such form too 
IIRC.

... because, I would not like to see following happen - you design 
some quick solution, which will not cover past VID missing part, 
which will not be easilly addable, and surely you will not feel comfort, 
if ppl will complain. We were there, and the response was - well, 
VID was written in about a week, and why you don't write your own 
one GUI engine upon View?

well, re-reading my posts - it can sound quite assertive, while I 
don't regard it being so. In fact I try to voice my opinion loudly, 
to prevent some past mistakes.

why you don't write your own one GUI engine upon View?

 ... a valid question, considering a "one size fits all" GUI is hard, 
 if not impossible, to do. I think [small] domain-specific GUI's built 
 on View are the way to go.

Ashley, I might know what you mean, but there should be one more 
complete/robust one, which will serve for "general" app development 
- simply put - most common styles behaving in OS compatible way, 
so that developer might feel safe to push rebol based apps around. 
Of course I can imagine specialised dialected UIs for presentations, 
multimedia, etc.

RT outlined VID 2 group - frankly I have never heard of such a thing 
from Carl.

well, who started the GUI Design group?

i started to talk of committee - no, and i'm sorry if i was not clear. 
we need help with the implementation, and we surely need feedback 
with the design (although, i think that we got quite a lot of feedback 
on this in the past few years, and from rebgui, liquidgl, and so 
on), but i don't think i ever side that design decisions should go 
thru a voting process or something like that. design decisions come 
from carl. i can show him my, or anyone else ideas, and he will say 
yes or no (i'm simplifying but you get the idea). showing him 100 
different ideas is not going to work (we must get to something in 
much less than a month)

hum, the "much less than a month" seems quite a bad idea in all ways 
I can approach it.

dataflow: me, carl, richard, nenad, max etc were sitting discussing 
this at the devcon, and nenad asked - i don't see what this is good 
for in practice. carl said, i'd just like the gui part of the thing 
- connecting widgets, so how big is liquid? max said at least 20k, 
to which carl said "too big".

if its going to take a year or two to properly adapt just a few small 
high-level things in the language... I don't see why redesiging view 
should be done in less than a month.

 I think [small] domain-specific GUI's built on View are the way to 
 go.

 -- I agree 100%. Better for RT to give us good docs on how to build 
 those, than trying to build them all in house.


but there should be one more complete/robust one, which will serve 
for 

general" app development" -- There is. Today it's called VID. It's 
not perfect, can be very limiting, and has big holes (e.g. no focus), 
but it's there.


I think the point Gabriele made--that this is how Carl works, and 
we have to live with it--is very important. That's not an easy thing 
to do, but I think getting too many people involved will not work. 
I hope there is a small team of technical people, and that someone 
makes note of what has been requested. I don't expect perfection, 
but I don't think RT will ignore what has been said in the past either.

so, i've looked into the liquid source, i've looked into what max's 
demos etc., and i still can't find an example where we need 20k of 
code. i can do all of that in 2k of code (actually zero, since i 
already want it to be all event based). so, i'm not making a final 
decision here, but i expect carl to chose 0kb over 20kb, functionality 
being the same from his pov (connecting widgets automatically).

but R3 is about finally allowing REBOL to DO real stuff.

Carl talks a lot about programming the large. I hope this aspect 
fits with the GUI as well, so if we can't get a complete GUI, at 
least give us the solid foundation to build it ourselves.

without that nagging bad taste in your mouth every time you try to 
do something where you realise you have to "again" reimplement half 
of what you are working on before even starting.

Agreed Henrik. That is my hope as well.

there are no Dataflow demos out yet.

liquid is not related to GUIs.

comments - we've been getting comments from here (and other various 
private groups) for the past two years or so... now we got to create 
it. i'll work as much as possible so that you guys have docs as soon 
as possible and before things are finalized, but we have to get it 
finalized by 30 june.

Gabriele, noone was thinking commitee aproach here imo :-) But IIRC 
we really talked about forming small group for VID+ era. I can't 
see anything bad on it. And I don't want to be there, because I know 
ppl here, I voiced what I need and I believe guys like Henrik, who 
are doing real-world apps know what we need in that respect ...

The other problem, at the risk of repeating myself, is that we all 
have different needs, so RT is never going to make all of us happy 
no matter what they do.

Gregg: exactly.

Would it hurt to invite those ppl to special World? :-) Or do you 
have enough input thru the years already, that you keep all our wishes 
in mind?

but if we can reuse some of the stuff without fighting our way through 
the code... it'll be godd for all.

(good)

my wish is simple - I want VID like environment, with redesigned 
problematic parts - simply put, it should allow OS like apps creations. 
I don't mind skinning at all, but things like focus, tabbing, accelerator 
keys, disabling/enabling of elements, etc. And probably more general 
VID level event system on-* handlers instead of one engage func.

I hope RT will get input from both Henrik and Ashley because of their 
experience in building on VID and View. More importantly, both of 
them have provided *great* docs.

if we make a new world to discuss this... is this getting us something 
new that has been said / discussed in the past two years or so?

Maxim - well, your stuff is so special, that I don't mind it having 
as a complete separate package, really. The same goes for RebGUI. 
I am ok with that. But I lost 2 ppl because of VID incompletness. 
VID is easy on surface, but difficult for ppl to extend with missing 
concepts - it would mean nearly a rewrite ... THAT is the factor 
I would like to avoid with new VID ... please :-)

well, Ashely actually decided to dump VID.

also, is funny how all this started because i said i wanted to have 
more people involved.

pekr: you mean like GLayout did?  ;-)

:-) I got the that too Gabriele.

not sure - for me it started when it started to be clear that you 
are going to do it in-house, behind the closed door, showing us the 
result only ;-)

Something I sometimes do for RT is collect and sift community input 
for them analyze. I donh't know if they always read it, but I know 
they do sometimes. If people have input they think is important, 
I will be happy to collect it and submit it to RT.

I would prefer that RT would work on the foundation, rather than 
something quick and sloppy that tries barely to cover everything 
like VID does. We'll help with the upper layers of GUI element design. 
That's not design by commitee, but simply compartmentalizing who 
does what. With R3, I had expected the goal to have as small and 
efficient a core as possible, with the rest being open source. Carry 
that philosophy through with the GUI as well, so RT can do a fast 
and efficient basis for a GUI and let actual artists and GUI designers 
work on the GUI.


This way, if some of us want a serious GUI system, we can build that 
and if we want something very fancy and artistic, we can build that 
too and both will not compete with eachother, but supplement eachother 
instead. I hate to see double work done in such a small community.

you know guys, in a sense I already have a 100% working dataflow 
view.  and it took me 2 hours to build the whole gadget architecture 
on it and about 30 minutes to write my first integer field... and 
its all AGG.  and its 100% bug free.

henrik, that is the plan, but the foundation has to provide a bit 
more than just "hey, you have gobs, you can make whatever you want 
with that".

I agree Henrik, with the exception that I want something like VID 
built in, something simple I know will always be there.

but its different... all it would need is a layout dialect (using 
GLayout resizing)

my problem with this situation is that you are judging us before 
we even started doing anything.

I'm using it in elixir, but AGG is a limiting factor right now... 
once it gets to pretty it starts slowing down.

so I really hope gobs will improve this :-)

Sounds very cool Max, is it posted somewhere, or wil it be?

cause all I'll need to do for R3 is replace my internal gob class 
(gel) with them and maybe how I detect the strokes and all.