That's technically pretty cool, but it makes me wonder:
In order to run a Java Desktop app, I need to install a JVM first (or the Desktop app can embed it, I guess that's what IntelliJ does, right?).
Now if I run CheerpJ, it means that I essentially download a JVM when I load the page (every time), and run code in that JVM. But at this point, why not downloading a Desktop app?
It feels like we are going around, shipping simple web pages together with full browsers and calling that "desktop apps" (e.g. ElectronJS), then shipping complete JVMs as web pages and calling that a "web page"... why don't we just ship simple webpages through browsers and complex desktop apps through package managers?
apignotti•10mo ago
With CheerpJ you are downloading the subset of the JVM that you need, and actually only once thanks to the standard browser cache.
There are many reasons why shipping via the browser is a better choice compared to shipping desktop apps. The main 3 in my opinion are:
1. Distribution: Give your user a link and the app will start
2. Isolation: The user can have confidence the app won't read his personal files.
3. Cross-platform: Every OS and every device, for real this time
yuri91•10mo ago
For reference, when loading https://browsercraft.cheerpj.com for the first time (up to loading a world), my browser downloaded ~32MB.
The second time almost nothing.
jeffreportmill1•10mo ago
And here's an entire Java IDE with CheerpJ that downloads less than 15mb:
> With CheerpJ you are downloading the subset of the JVM that you need
That's interesting! May I ask how it works? Does that also happen with e.g. IntelliJ?
> Every OS and every device, for real this time
Doesn't the JVM run everywhere in 2025?
apignotti•10mo ago
> That's interesting! May I ask how it works? Does that also happen with e.g. IntelliJ?
Byte ranges request do most of the heavy lifting, data is loading exclusively on-demand.
> Doesn't the JVM run everywhere in 2025?
What about iOS? Android has Java, but can't run desktop Java apps. Chromebooks also have limits.
palata•10mo ago
> Byte ranges request do most of the heavy lifting, data is loading exclusively on-demand.
I don't understand what that means. The JVM is supposed to interpret and sometimes compile bytecode, right? How can it be done with only a fraction of the JVM?
Or are you saying that it is constantly communicating with a server that does the work?
apignotti•10mo ago
The VM itself is very small, it's the OpenJDK runtime that is quite sizeable. Byte ranges are used to only download the parts of the runtime (in terms of bytecode) that are required.
There is no server-side computation. CheerpJ runs code exclusively client-side.
palata•10mo ago
But you said before that you only download a subset of the JVM, right? Or did you mean a subset of the JDK, including the JVM and... I guess other stuff?
apignotti•10mo ago
I meant the JVM in an extended sense: the combination of the bytecode parsing, JIT compiler and OpenJDK runtime. You are right, I should have been more precise and refer to only the runtime part, which is by far the most significant.
palata•10mo ago
I was not trying to prove you wrong, I'm just genuinely interested :-). I don't see a lot of articles about the JVM these days.
palata•10mo ago
In order to run a Java Desktop app, I need to install a JVM first (or the Desktop app can embed it, I guess that's what IntelliJ does, right?).
Now if I run CheerpJ, it means that I essentially download a JVM when I load the page (every time), and run code in that JVM. But at this point, why not downloading a Desktop app?
It feels like we are going around, shipping simple web pages together with full browsers and calling that "desktop apps" (e.g. ElectronJS), then shipping complete JVMs as web pages and calling that a "web page"... why don't we just ship simple webpages through browsers and complex desktop apps through package managers?
apignotti•10mo ago
There are many reasons why shipping via the browser is a better choice compared to shipping desktop apps. The main 3 in my opinion are:
1. Distribution: Give your user a link and the app will start 2. Isolation: The user can have confidence the app won't read his personal files. 3. Cross-platform: Every OS and every device, for real this time
yuri91•10mo ago
The second time almost nothing.
jeffreportmill1•10mo ago
https://reportmill.com/SnapCode
palata•10mo ago
That's interesting! May I ask how it works? Does that also happen with e.g. IntelliJ?
> Every OS and every device, for real this time
Doesn't the JVM run everywhere in 2025?
apignotti•10mo ago
Byte ranges request do most of the heavy lifting, data is loading exclusively on-demand.
> Doesn't the JVM run everywhere in 2025?
What about iOS? Android has Java, but can't run desktop Java apps. Chromebooks also have limits.
palata•10mo ago
I don't understand what that means. The JVM is supposed to interpret and sometimes compile bytecode, right? How can it be done with only a fraction of the JVM?
Or are you saying that it is constantly communicating with a server that does the work?
apignotti•10mo ago
There is no server-side computation. CheerpJ runs code exclusively client-side.
palata•10mo ago
apignotti•10mo ago
palata•10mo ago