Page 2 of 6

Re: Successful test

Posted: Mon Dec 14, 2015 11:34 am
by Dixie
Hi Hermann..

You are quite right... The stack I sent does not run..:-( So, I got out the original stack that I tried earlier this morning, opened it in LC8(dp11) and then made an HTML5 standalone from that... It runs, along with the stutter..:-)

El Capitan 10.11.1, liveCode 8(dp11) community

I have attached the stack..

Re: Successful test

Posted: Mon Dec 14, 2015 12:22 pm
by Dixie
Shall I host the standalone of you stack with my collection?
If you wish !... I will try and find some time and look at the scripts, to see if it can be improved...

Re: Successful test

Posted: Mon Dec 14, 2015 1:51 pm
by [-hh]
We will wait, also for your next highlights.

Until that we can try John's "Link 4" HERE (extern server) or from the overview of the 'community-page' of my experiments collection.
dixie wrote:If you try it... 'click' on the logo in the bottom left of the stack to start a new game.

Re: Successful test

Posted: Thu Dec 17, 2015 9:57 am
by peter-b
Hi Hermann,

Are you experiencing the bugs in Commercial or Community edition of DP 11?

When I made the changes for making "wait" work, I accidentally broke Commercial edition for HTML5 -- but Community edition should work just as well or even better than DP 10.

Re: Successful test

Posted: Fri Dec 18, 2015 10:02 am
by peter-b
[-hh] wrote:Now all my dp-9 examples run also in community-dp-11. I will update soon all that don't "slow down".

"Slow down" is an issue that arose with 'animating' in dp-11, the amount is at about the same as the speed amount lost from LC 6 to LC 7:
In HTML5 standalones the animation by 'move' isn't smooth any more, changed to kind of "stuttering", looks like screen update frequency is changed to 10-15% of the previous. [Edit: Reported for that (QC #16619)].
A good example for that is the HTML5 demo of the game "Link 4" (by 'Dixie', animation is 'soft-smooth' in the IDE), see here. He uses to 'set the movespeed', followed by a plain 'move' and closing with some 'wait' interval. I use always 'move <in time>'.
Argh. So, this is the fundamental trade-off. The changes made in 8.0.0-dp-11 in order to enable "wait" to work were to make much more of the engine be run using a special JavaScript-like byte code that, unlike "proper" JavaScript, can be paused and resumed. Unfortunately, this byte code is also much slower to run. So, "wait" works now, but the engine is slower.

What's really necessary is to basically rewrite the whole way that LiveCode script evaluation works, so that the whole engine can be run as native JavaScript. Unfortunately, I cannot do that before 8.0.0 release (actually, I would be surprised if I can have the resources to do it before LiveCode 9 is released, even though it would be a very big improvement on all platforms, not just HTML5!)

I think it will be necessary to have the slower engine for LiveCode 8.0.0, because otherwise I cannot make the "URL" syntax work.
[-hh] wrote:p.s. Will you update the Indy version of 8.0.0-dp11?
The Indy and Business updates will be in LiveCode 8.0.0-dp-12, which I think is expected to be built today.

Re: Successful test

Posted: Mon Dec 21, 2015 10:58 pm
by [-hh]
Tested now all my examples above and some of others with dp11-dp13, everybody can replicate this by himself with the downloadable stacks.

This is just to inform you about the result.
  • The 'slowdown' is so extreme that even medium slow typing in a text field 'stutters'.
  • 'Move' and 'grab' work but unregularly at *very* low speed.
    I measured a factor of 10-40 of increase in needed time compared to dp9 (which is at about at RaspberryPi B speed -- 700 MHz).
  • Sliders work close to a fly-by-night.
That is in sum (more exactly in product) a factor of 100-400 of increase in needed time compared to the IDE on a medium fast machine, say 2.5 GHz.

That is:
Who is speaking about milliseconds on such a machine speaks about tenths of seconds on the fastest available browser, cross-OS and cross-Browser problems not included.

:( :(

Nevertheless, Peter and the team:
Thanks for your work in 2015 and have nice holidays until 2016.

I hope you'll find enough time to work on this next year.

Re: Successful test

Posted: Wed Jan 06, 2016 11:28 pm
by peter-b
[-hh] wrote:We were promised, that at latest the golden-master-version 8.0.0-gm is fast enough to blow up our kilts (the ones we bought in Edinburgh).
I'm pretty sure I didn't promise that?!

I'll do my best though. :wink:

Re: Successful test

Posted: Thu Feb 25, 2016 2:34 am
by [-hh]
Whoever wishes to try standalone examples may have a look
HERE or (same content, different server) HERE.

More testing by community members.
Richard's WebApps
Malte's AE first-test
Bernd's "little webapps"
Dixie's examples

With CountUpOrDown (default target date is Aug 2, 2016, 08:42), watch especially the
loading animation: LiveCode logo showing progress percentage!

Currently, using LC 8.0.0-dp15 for generating the standalones,
we have with most animations the following running speed.
LC IDE = 1 time unit (=U) as base and fastest:
Firefox 2 U, Safari 4 U, Chrome 7 U, Opera 8 U.
The loading speed of the HTML standalones is the same ranking from fastest (left) to slowest (right), but samller differences.
The 'slower' browsers have other display advantages than speed, giving different comparison order.

Re: Successful test

Posted: Thu Mar 03, 2016 1:08 am
by [-hh]

My input stacks for the HTML5 standalones (made with a commercial license) are freely available for the community. The scripts are not password protected.

You can use it or parts of it freely if and only if for **non-commercial** projects.

When using special 'signed' script parts or groups like my colour choosers, you could cite me: "LC community member [-hh]" is OK for that.

Where to get?  Simply replace "X.html" (or* "hhX.html", see the url) by ".zip". Have fun!
Actual Overview HERE or HERE.
* "hhX.html" is used as ending whenever the webpage loads also (needed) js-files.


p.s. Note. Most of these stacks are simplified versions of stacks of the "RaspberryPi stacks collection" (simplified to have them running with the HTML5-alpha versions).

Re: Successful test

Posted: Fri May 06, 2016 1:44 am
by [-hh]
Newest Mathy Beauties:

Mathy Beauties 2 (Neon) --- Animations of "balls" (circle graphics) along closed curves.
Several parameters (curve, size, speed, numOfBalls) may be set.
You may have to adjust things depending on your CPU speed and web browser.

Look at it frome here (EU) or from here (US).

This is my first test with the new LC 8.0.0 stable release. It works pretty good (as nothing essential has changed since dp-15).

[The animation there is always 'in sync' of every 2 ticks (='best' browser refresh rate). "Speed" comes from selecting less than every single vertix of the closed curve the points move along.]
* serves now from North-East US.

Re: Successful test

Posted: Fri May 06, 2016 5:20 am
by Dixie
Herman... that is excellent !...:-)

Re: Successful test

Posted: Tue May 10, 2016 12:06 am
by [-hh]
[Since John certainly doesn't like every HTML5-modul I make, I have to start a new post. Thanks John, for the flowers.]

Powers-Of-Two Clock as HTML5 demo.

This is perhaps one of the most complicated 'mathy' clocks I made and at the same time one of the simplest, so this needs some explanation.

For the clock's display the date and time's *single* digits are first coded into powers of two. That is: 18 is coded into 2^1 and 2^8, 27 into 2^2 and 2^7 and so on.
Now we use any number base out of 2-60 for the display of these numbers.
That is, if we use base 10:

18 -> (2^1 and 2^8) -> (2 and 256)
27 -> (2^2 and 2^7) -> (4 and 128)

so that 18:27 -> (2 256) : (4 128) ... (read the 'and' as space).
For any other number base out of 2-60 we simply encode these last decimal numbers into that base.

For example, if we use base 9:
18 -> (2^1 and 2^8) -> (2 and 256 -> (2 and 314) (= 3*9^2 + 1*9^1 + 4)
27 -> (2^2 and 2^7) -> (4 and 128 -> (4 and 152) (= 1*9^2 + 5*9^1 + 2)

Vice versa, when reading the display, we take for simplicity base 10 as intermediate.
First we convert the number to base 10 and take then the log2 to have the original digit, for example (read 'space' for 'and'):
(4 and 152) at base 9 -> (4 and 128) at base 10 -> ( log2(4) and log2(128)) = (2 and 7) = 27.

Sounds complicated, but it's not:
Take as base b=2. Then watch the display and tell the time. Yes you read from PowerOfTwo-coded-decimals displayed at base 2. Be sure, you can read the clock display at once! This base choice gives one of the simplest clock displays I ever made.
("It's funny", said my little nephew, "zero zeroes means zero!")

Look at it frome here (EU) or from here (US).

Re: Successful test

Posted: Sun May 29, 2016 9:53 pm
by [-hh]
Wireframe Fonts
Tracing of text: contour, horizontal, vertical, gutter or points.

This is a 'byproduct' of the preparation of my talk in August (Edinburgh).
Just noticed some nice effects we can have using LC's display options for polygons.

Look at it from here (EU) or from here* (US).

If you want to have this fast (10-15 times faster) or use your own fonts as 'base' then you may use the IDE version with LC 6, available in the RasperryPi stacks collection.

* serves now from North-East US.

Re: Successful test

Posted: Tue Aug 16, 2016 4:44 pm
by [-hh]
LCD Scroller
(Creating and serving a HTML5 standalone)

The PDF (the 'slides') of my talk at the LC conference, Edinburgh 2016.
The resulting standalone may be viewed here (EU) or here (US).

If you want to have this fast (2-10 times faster) then you may use the IDE version with LC 6, (soon) available in the RasperryPi stacks collection.

Re: Successful test

Posted: Wed Aug 17, 2016 5:32 am
by Richard K. Herz
Beautiful work Hermann! Thank you for this stack and your talk at LiveCode 2016!