World: r4wp

There are two main types of documentation, reference docs and guides

I'm looking forward to making Red versions of my tutorials and examples 
scripts.  Those will be guides - someone else will need to focus 
on reference docs.

I am thinking about a reference documentation. How to do it without 
rewritting REBOL/Core documentation?

If Carl is going to open source the R3/3 documentation,
 I haven't read anything about that.

I have some cool ideas about guides too, mainly for first steps with 
Red and how to clearly explaining concepts to beginners, but I have 
issue imagining a reference documentation content different from 
REBOL/Core docs.

So, can we do a better than REBOL/Core documentation? Can we present 
the same content in a radically different (better?) way?

(By REBOL/Core documentation, I'm referring to the "old Core manual" 
as Gregg said)

About the tooling, a wiki approach + makedoc format makes sense to 
me.

Well, as for me, it is easy as that - f*ck off native controls :-) 
What does that really mean today? I have really no intention seeing 
GTK or other crap. Either use your own GUI, or HTML, that's almost 
the only standard, which will prevail anyway. We should ask ourselves 
- how relevant is e.g. the Gnome GUI guide? And who cares in the 
big picture? So - for easy embedded stuff, just give me a small GUI 
system, for bigger picture, give me a  HTML5. I e.g. don't understand, 
why Kaj tries to port something like Enlightenment. I really miss 
the big picture. Who and why should ever use it? For personal purposes, 
or some specific stuff - why not, but that's not imo a strategic 
direction ...

I do remember the times when Win95 came. All the world was claiming 
- your apps need the menu, a toolbar, etc. Free style Amiga like 
apps were a dirty world. Now, 15 years later, such aproach looks 
archaic, laughable. The time has changed. Let's target mobile platforms.

And I even don't agree with Henrik. I really can't see, how your 
top-down aproach might work. You need a solig gfx engine (View), 
general enough, to build up. Carl's GUI was OK. And imo Saphirion 
did a bad mistake - we heard, for so long time, that the look is 
the final step. All those years, and the look is really a crap. Much 
worse, than what Carl brought up, even if I can see many improvements 
in engine itself. Look sells, take it, or leave it, and then - please 
don't even try to do your own GUI. No matter how good it is, if it 
looks like 80'ties Solaris, it will never get accepted ...

Pekr, the look is only crap, because I don't have time to work on 
the skin yet.

So you're saying don't use native toolkits and don't do your own 
GUI? ;-)

And appearance really is quite separate from the underpinnings.

No, it is just how you understand it. After 15 years, I still can 
see similar mistakes ....

I would never want my app to look like GTK, easy as that. Maybe because 
I have zero intention to do Linux native GUI app. Linux, in a big 
picture (desktop), is irrelevant. Alway was. Who claims otherwise, 
lives in separate reality :-) Mobile will prevail classical PC sales 
very soon (if not already). Own GUI, for small niche, might be good. 
But - it has to have attractive design. The situation with Saphirion 
was, that the substantially rewritten Carl's engine. Nice enhancements, 
resizing and other stuff. But I NEVER understood, why the look was 
so totally unattractive. You see, Henrik once again claiming the 
old stuff, which was the basic reason of a failure. Design is a separate 
animal, which was supposed to come later, which could be done by 
anyone, which "just" needs material system which never appeared, 
etc. And result for me (as a sigle person)? Carl's GUI desing attracts 
me by some 60% more, just because of look. Once again - look sells 
...

In fact, what I think is, that Saphirion was really close to the 
GUI engine, which adressed most of what we wanted to solve by VID2 
to VID3 transition. It was just not polished. And because of that, 
ppl did not find it attractive enough to play with. And that's a 
real pity. Anyone doing native platform GUI will make me to close 
that demo at first sight, easy as that. Doc is in kind of difficult 
situation - as we can see, many former/recent rebollers are still 
interested in View like engine. The same reason why Doc dismissed 
LLVM in Carl's blog post, the same reason many will dismiss attempts 
to link to GTK, Enlightenment, etc - I don't want to use 5+MB crap 
...

which 

just" needs material system which never appeared, etc." - the material 
system should be included in the current Saphirion release.

...for easy embedded stuff, just give me a small GUI system, for 
bigger picture, give me a  HTML5


So what do you propose for the GUI, that will allow it to look good? 
i.e., what is your constructive suggestion, rather than criticizing 
all the wrongs of the past.

Petr, if you pay me enough, I might refrain from doing the GTK+ and 
Enlightenment bindings :-)

Gregg - my constructive suggestion is to orientiate on the platforms, 
which have a future. Is that an AmigaOS (to which R3 got ported as 
a paradox), is that a BeOS? Is that even a Linux? I would vote - 
Android, iOS, WP8, BB10. If new players with new potential appear, 
let's add them. How many of us would ever need to do a real native 
Linux app? And Linxu, is a big name here.

So - it might seem, that I contradict what Doc wants. I think I don't 
do so. He wants to be native on those above platform. And that makes 
sense. But - are those above platforms done in GTK? Enlightenment? 
Look at Microsfot - I just yesterday was watching their 5 hours of 
SharePoint plus Office 2013 presentation - most of their new stuff 
is becoming - HTML5. They are even scratching Silverlight for the 
typical desktop ...

My suggestion, then, is to make a list of target OSs, prioritize 
the value of a GUI on them, and propose the best GUI solution for 
each.

Would you be willing to set up a checklist for that Petr?

Pekr: you don't get the right picture. When I mention "native widgets", 
I am mainly referring to the ones provided by each OS, which is what 
most users expect to find in an app for their OS, and what most developers 
wants, is to provide a consistent experience for users. So, actually, 
such approach will be lighter then /View, because the OS provides 
you with everything you need. In the case of Linux, GTK is the main 
standard and it is built-in many distros, so that is the one we will 
probably use for Linux target and you don't need to provide it with 
your app.

Or even just a shared doc here.

I think the check list is done :-) I think that in 2-3 months, Doc 
wants to start by - Android :-) And maybe something in that direction 
is already slowly starting to happen :-)

Where is the checklist?

No checklist, just general wishlist ...

Doc, fine - then I'll never use Linux as a desktop - hate it, easy 
as that, so no REBOL/Red apps for me there  ... What I wonder even 
more is, if you would be really able to unify, even in the case of 
Linux - what about e.g. Ubunty, and their Unity? Will it work just 
with GTK? Will such app look native in Ubuntu? Just curious ...

GTK is the native GUI for Ubuntu. I've told you many times

How many of us would ever need to do a real native Linux app?


We are talking about a cross-platform GUI framework, that means: 
write once, run everywhere. Using native widgets from the OS is an 
option, using custom ones (with non-standard look'n feel like /View) 
is another option. So, you write a Red GUI app on whatever system 
you like and your Linux users will be able to run it.

Unless you want to boycot Linux users and prevent them from using 
your Red-based product because you don't like Linux? :-)

Yes, that's an old discussion. In case I want to be crossplaform, 
I only accept it with mobile OSes, which are very special. I want 
my View app, I want to look and behave it identical on all plaforms. 
Can't imagine how I would be using eg. this Altme on Linux, using 
GTK, then switching to Win7, having totally differnet look, feel, 
behaviour in details ...

Feel free to write a custom GUI in Red

Linux is dead since its beginning, as far as desktop goes anyway. 
It is here since mostly 1994? Where's the world domination? Look, 
I welcome alternatives, I grew up with them, I am just realistic. 
E.g. Haiku has some drive, so much energy of so many ppl spent, but 
hardly becoming a mainstream.

It's hard to imagine that the native UI interface would meet with 
criticism from any main stream developers - by definition, it's the 
_native UI.  Doing things that way just takes longer to port.  I 
honestly don't care which way it goes, as long as there's a usable 
GUI system with a _productive syntax_.

So, you want something like VID/View, that looks great, and works 
on mobile platforms. Is that correct Petr?

Kaj - I feel free to find some resources for someone to port View 
to Red, playing on your nerves, as it will quickly again becoming 
more popular :-)

Gregg - no, I want ALSO View/VID, which is being kind of dismissed 
here both by Doc and Kaj :-)

My nerves are fine. It's yours I'm converned about :-)

How is that different from what I said (and I think they agree that 
you can write it if you want)?

I tried to attract Cyphre, unfortunatelly he has new View like engine 
in mind, which would require some extensive resources to spend to 
get that. And I don't have such resources, so I am "a bit" nervous 
:-)

That is, if Doc creates your dream GUI system, what is it?

I should have asked the GUI question in private ;)

:-)

No problem, Petr will complain no matter what

Doc starts with a kind of dialect - some midd level. Kind of what 
Henrik called top-down, but with good thinking ahead. you mostly 
write VID-like dialect, which would get (not necessarily 100% automatically) 
compiled to target platforms. So on Android, it looks like andorid, 
etc., but also - for small embedded - either html5, or View ...

My understanding is that most people stick to one desktop OS, they 
are not switching OSes whole day (that is just a very small minority 
of computer users). Having close integration with OS can bring many 
advantages over a custom GUI like hardware acceleration or close 
integration with OS/desktop services. There are still cases where 
a custom cross-platform GUI could be good enough, so a View-like 
engine for Red is welcome. I just think that the default GUI option 
should be to provide an OS-integrated user experience.

Ah, I thought so, could count minutes before someone comes with "hardware 
acceleration claim". We havent ever utilised full speed of new View 
engine in R3, yet we want it even faster :-) But understood, some 
mobile devices might not have FPU or so ...

So on Android, it looks like andorid, etc., but also - for small 
embedded - either html5, or View ...

That sounds like what Doc is proposing.