World: r4wp

I was beginning to despair, but I finally got the Enlightenment canvas 
library working on Syllable Desktop :-)

I got some of the examples of the Edje layout engine to work

Kaj, not sure if it's of use to you, but I dug out the timings from 
the last time I built Qt (on a Linux host, 4 months ago):


$ ./configure -opensource -nomake examples -nomake demos -nomake 
docs -nomake translations
5 minutes

$ make -j2
34 minutes


That was on a moderate dual-core machine with 2GB ram and spinning 
disks. Not sure how much make could parallelise, but for a single-core 
machine roughly twice the build time will be a reasonable worst case 
estimation. I guess that otherwise the configure-to-build time ratio 
will still be roughly the same.

(Sorry, ~2 months ago.)

I'm starting to get the GUI build to work, and it takes a few hours 
here, but Syllable Desktop can't use the second core on this machine, 
and compiling is roughly half the speed as on Linux, anyway, so that's 
roughly consistent with your timing

The configuration can be sped up a bit by using -fast if you are 
using -nomake like that on significant parts

The QtQUI shared library I have right now is 12.5 MB

Yes, ~13M for the libQtGui shlib here as well.

The installed QMake I end up with for Qt 4.8.2 is almost 5 MB

That's good footprint, isn't it? Not following the discussion - I 
thought you are working on an Enlightenment, not Qt? But maybe I 
mix Syllable and RED efforts?

I don't know what you think the footprint is for, but it's only the 
executable of one tool needed for compiling Qt

ah ha, I thought it is a "runtime" library with the whole Qt :-)

No, that's around 100 MB, I have now established, if you strip out 
all the documentation, examples and demos

So Syllable is going to use Qt for some apps or the system in overall?

So what's your Enlightenment effort then? Red only related?

No, we never wanted to depend on Qt. And I have it all built now, 
but the GUI doesn't work yet

Still, people want to run Qt apps

I'm working on many things, currently Enlightenment, GTK+ and Qt. 
The infrastructure they need in Syllable to build and run is mostly 
shared, and some of the dependencies they use are shared

I have all of Enlightenment working except the widget set. I have 
DirectFB working and am currently building GTK and Qt on top of that. 
I have all of Qt built but nothing GUI works. I have all of GTK working 
except GTK itself, so no GUI there, either, but I'm still working 
on that

You seem to be skilled in porting various toolkits. No intention 
to port View engine to Red/System? :-)

As Gabriele noted, I'm modifying code I don't understand, so I'm 
not feeling very skilled, but I suppose I am after the experience 
:-)

It doesn't help that the ports get stuck just before they would become 
useful - as has happened with many ports attempts over the years

So I need very good reasons to start such an attempt, and the R3 
View engine has several red flags that we discussed before

I was going to port it to Syllable, but with REBOL down that makes 
no sense anymore

what red flags? The licence is not set, and the author and owner 
of the code is Cyphre, not RT, so if it is licence related, we should 
talk to Cyphre imo ... but anyway - wrong group. It is just that 
we are binding to some heavy solutions as GTK, Enlightenment, instead 
of going probably the less hard route?

Those toolkits have widget sets that you would have to redo in Red 
for View, for another several years of uncertain development

Even Cyphre doesn't want to use it anymore, for technical reasons

I know. But I also expect RED being kind of compatible to RED, syntax 
wise. And I think Robert would not mind, if someone tried to port 
R3 GUI to RED. As you can read on various places, ppl still want 
View, small toolkit, and some are even reluctant to join RED, if 
the clarification of GUI availability is not made ...

Even Cyphre doesn't want to use it anymore, for technical reasons

 - I never heard anything like that, and it even does not correspond 
 with my info, that in fact Cyphre would like to redo the View engine 
 completly ...

I have already provided a GUI for Red. It's fine if R3/View is rewritten 
for it, but I'm not going to do it

Well, Ok, I need to get rich, to pay some dev to port View to RED 
for my purposes :-)

What's your GUI for RED? GTK? How big is that? No under 1MB exe, 
no sugar :-)

I'm porting those toolkits to Syllable now exactly to clarify which 
options I can support

Cyphre redoing the View engine is the same as not wanting to use 
the current one anymore, isn't it?

No, it was just his long term idea, to abstract the engine, so that 
it can use various backends - AGG, Cairo, etc, and provide platform 
acceleration, where available. Unfortunatelly, that was just an idea 
on his side, no real project. There is nothing wrong with View engine 
itself, apart from some bugs in core, which make it look unfinished. 
That would not happen with Red, as all sources and hence debugging 
is possible ...

If you port it to Red, you'll only have a drawing engine, no widgets. 
I already have that in my Enlightenment port to Syllable, or in SDL 
or DirectFB if you combine it with some drawing library

But other engines you name are far from what View engine in fact 
is - system of gobs, etc. Widgets are VID. I still prefer not so 
much perfect VID, instead of overbloated stuff ...

As I've reported before, Enlightenment is quite like View regarding 
the architecture

How big are the Enlightenment libs? Does E have any peculiar build/runtime 
dependencies?

It's a pretty integrated stack, but modularised into more than ten 
packages. The dependencies are pretty much the same as the basic 
dependencies for other toolkits such as GTK, and most of them are 
optional. Stuff such as FreeType and FontConfig:

http://www.enlightenment.org/p.php?p=download&l=en

Very roughly speaking, when you add up the libraries, Enlightenment 
is half the size of GTK which is half the size of Qt

The shared libraries on my system currently total 5 MB, without dependencies 
from outside Enlightenment

Thanks.

Cyphre redoing the View engine is the same as not wanting to use 
the current one anymore, isn't it?

 - Well, I actually use the current one with R3 almost everyday ;-) 
  And to make things clear I won't be redoing it...my plan is to update 
 it to 'next generation'  also add more features and improve some 
 parts of the current base etc.

I thought you found AGG too slow on your phone?

Kaj - I do remember View 1.0 alpha with CID (predecessor to VID) 
on a Pentium 75, 130 - ran "acceptable". Cell phones have limited 
UI needs imo, I doubt AGG will be slow. Of course some heavy operations 
might drag some juice from the battery, as it is not accelerated. 
We now need to find the ways of how to get Cyphre's idea becoming 
a reality ...

The issue is with phones that don't have FPUs. I'm just basing myself 
on what Cyphre said several months ago

Kaj, yes, but the changes I plan will allow you to relatively easily 
use different renderer component. Even in current host-kit I would 
just replace the agg renderer with something more suitable for slow 
or not sufficiently equipped  ARM cpus but the "framework base" of 
the sytem would remain same.

Pekr, my experiment showed AGG on ARM without FPU was way slower 
than the native implementation of Android Canvas engine...mainly 
because the Canvas uses integer based rasterizer etc.