Re: Talk on Royale

[Roman Isitua]

I have a question to ask regarding the load time of the spectrum browser
(it's really fast) when compared to the demo spectrum app I deployed
online. I will ask my question on another thread shortly.

On Wed, Jan 26, 2022 at 1:34 PM Harbs  wrote:

> Reaction was positive. Not sure I convinced anyone to switch, but the more
> word gets out, the better…
>
> I added some comments to give a better picture of performance gains. Not
> sure why it’s not showing up.
>
> Here’s what I added:
>
> I did not do a very good demonstrating the performance benefits of Roylae
> in the talk. I put some better numbers together:
>
> *Minimal app*
> js 17KB gzipped
> html 444 bytes unzipped
> html 270 bytes gzipped
> css ... depends on what you put there
> load times:
> 95 ms document load event
> 100 ms onload event
> 109 ms for FP, FCP and LCP
>
> So we’re talking less than 20KB of data TOTAL and about 100ms load time
> for a simple app.
>
> *Spectrum Browser*
> js 204 KB gzipped
> html 5,632 bytes unzipped
> html 863 bytes gzipped (with references to 64 external css files)
> load times
> 275 ms FP FCP
> 281 ms doc load
> 330 ms LCP
> 343 ms onload
>
> This *easily* beats Google’s optimal load times of less than 1/2 second.
> In fact you need to try VERY HARD to build a Royale app which takes longer
> than 1/2 second to load.
>
> Here’s a breakdown of the load time of the Spectrum Browser:
> The bulk of the load time is due to the loading of Adobe’s Spectrum CSS.
> Some of things I’d like to do is add a Semantic-UI component set and create
> a tutorial on using Tailwind CSS with Royale. A build process which
> generates optimized external CSS would do wonders for load times...
>
> The first 159 ms was evaluating css files. I’m loading 64 css files from
> Adobe, so that takes time. If I cared, this can probably be optimized a lot.
>
> Evaluating and building html dom tree took 94 ms. Of that, evaluating the
> script took 65 ms. Not sure why that took so long.
>
> The actual running of the code which builds the DOM took 27 ms.
>
> Recalculate style took 20 ms. If I decided that was too long, I could
> delay loading some of the content pretty easily which would speed this up.
>
> All in all, I don’t feel that a 204KB js file (with zero other
> dependencies) and a 300-350 ms load time is something that needs further
> optimization. If I did, Royale supports modules. I could probably use
> modules to cut down the JS size and the load times to less than half. Not
> worth it IMO.
>
> Like I mentioned in the talk, switching routes generally takes about 20ms.
> That’s about a frame rate of about 60 fps to switch routes. That’s insanely
> fast if you stop to think about it...
>
> I hope that does a better job of demonstrating the performance benefits of
> Royale.
>
> On Jan 26, 2022, at 9:11 AM, Yishay Weiss  wrote:
>
> Nice presentation. Too bad there wasn’t a Q&A. What was the reaction?
>
> *From: *Harbs 
> *Sent: *Tuesday, January 25, 2022 1:57 PM
> *To: *users@royale.apache.org
> *Subject: *Talk on Royale
>
>
> I recently gave a talk on Royale to a local group in Jerusalem. Please
> excuse the poor delivery. I was recovering from Covid and a flu. ;-)
>
> I went a bit into strands, beads and performance.
>
> I used the router classes as an example of MXML and Royale features.
>
> It might be of interest. There will be some colloquial terms you might not
> recognize, but it should all be understandable...
>
> https://www.youtube.com/watch?v=QWGISrmN4to
>
>
>

Re: Talk on Royale

[Roman Isitua]

This is nice. I have bookmarked this link and will watch it.

On Tue, Jan 25, 2022 at 12:56 PM Harbs  wrote:

> I recently gave a talk on Royale to a local group in Jerusalem. Please
> excuse the poor delivery. I was recovering from Covid and a flu. ;-)
>
> I went a bit into strands, beads and performance.
>
> I used the router classes as an example of MXML and Royale features.
>
> It might be of interest. There will be some colloquial terms you might not
> recognize, but it should all be understandable...
>
> https://www.youtube.com/watch?v=QWGISrmN4to
>
>
>

Re: Talk on Royale

[Harbs]

Reaction was positive. Not sure I convinced anyone to switch, but the more word 
gets out, the better…

I added some comments to give a better picture of performance gains. Not sure 
why it’s not showing up.

Here’s what I added:

I did not do a very good demonstrating the performance benefits of Roylae in 
the talk. I put some better numbers together:

Minimal app
js 17KB gzipped
html 444 bytes unzipped
html 270 bytes gzipped
css ... depends on what you put there
load times:
95 ms document load event
100 ms onload event
109 ms for FP, FCP and LCP

So we’re talking less than 20KB of data TOTAL and about 100ms load time for a 
simple app.

Spectrum Browser
js 204 KB gzipped
html 5,632 bytes unzipped
html 863 bytes gzipped (with references to 64 external css files)
load times
275 ms FP FCP
281 ms doc load
330 ms LCP
343 ms onload

This easily beats Google’s optimal load times of less than 1/2 second. In fact 
you need to try VERY HARD to build a Royale app which takes longer than 1/2 
second to load.

Here’s a breakdown of the load time of the Spectrum Browser:
The bulk of the load time is due to the loading of Adobe’s Spectrum CSS. Some 
of things I’d like to do is add a Semantic-UI component set and create a 
tutorial on using Tailwind CSS with Royale. A build process which generates 
optimized external CSS would do wonders for load times...

The first 159 ms was evaluating css files. I’m loading 64 css files from Adobe, 
so that takes time. If I cared, this can probably be optimized a lot.

Evaluating and building html dom tree took 94 ms. Of that, evaluating the 
script took 65 ms. Not sure why that took so long.

The actual running of the code which builds the DOM took 27 ms.

Recalculate style took 20 ms. If I decided that was too long, I could delay 
loading some of the content pretty easily which would speed this up.

All in all, I don’t feel that a 204KB js file (with zero other dependencies) 
and a 300-350 ms load time is something that needs further optimization. If I 
did, Royale supports modules. I could probably use modules to cut down the JS 
size and the load times to less than half. Not worth it IMO.

Like I mentioned in the talk, switching routes generally takes about 20ms. 
That’s about a frame rate of about 60 fps to switch routes. That’s insanely 
fast if you stop to think about it...

I hope that does a better job of demonstrating the performance benefits of 
Royale.

> On Jan 26, 2022, at 9:11 AM, Yishay Weiss  wrote:
> 
> Nice presentation. Too bad there wasn’t a Q&A. What was the reaction?
>  
> From: Harbs 
> Sent: Tuesday, January 25, 2022 1:57 PM
> To: users@royale.apache.org 
> Subject: Talk on Royale
>  
> I recently gave a talk on Royale to a local group in Jerusalem. Please excuse 
> the poor delivery. I was recovering from Covid and a flu. ;-)
> 
> I went a bit into strands, beads and performance.
> 
> I used the router classes as an example of MXML and Royale features.
> 
> It might be of interest. There will be some colloquial terms you might not 
> recognize, but it should all be understandable...
> 
> https://www.youtube.com/watch?v=QWGISrmN4to 
> 

RE: Talk on Royale

[Maria Jose Esteve]

Curious... 😝 
Were there any comments?

Hiedra

-Mensaje original-
De: Harbs  
Enviado el: martes, 25 de enero de 2022 12:57
Para: users@royale.apache.org
Asunto: Talk on Royale

I recently gave a talk on Royale to a local group in Jerusalem. Please excuse 
the poor delivery. I was recovering from Covid and a flu. ;-)

I went a bit into strands, beads and performance.

I used the router classes as an example of MXML and Royale features.

It might be of interest. There will be some colloquial terms you might not 
recognize, but it should all be understandable...

https://www.youtube.com/watch?v=QWGISrmN4to