Is there any way to provide missing implementations?
Nacho Cordón
- java.util.WeakHashMap: I can see a pr trying to add it was never accepted
- java.util.concurrent.CopyOnWriteArrayList
- java.io.InvalidClassException, java.util.CancellationException
- PrintStream.format, PrintStream.printf, String.format
Alexey Andreev
Hi
When we transpile some code with TeaVM we can see it's missing some implementations:
By saying "we" whom do you mean? Can you introduce yourself?
- java.util.WeakHashMap: I can see a pr trying to add it was never accepted
This one is really tough. One needs to understand how TeaVM virtual thread work under the hood and how to implement thread emulation code that would avoid async infection in single-thread scenario. Also, it's usually quite challenging to write tests properly.
- java.util.concurrent.CopyOnWriteArrayList
- java.io.InvalidClassException, java.util.CancellationException
These are trivial
- PrintStream.format, PrintStream.printf, String.format
I believe these could be implemented by simple delegation to the
functionality that is already implemented. I'm only not sure which
one of overloads you need, for example, for String.format.
Some of those I can maybe work on an implementation to add to TeaVM as a contribution if the maintainers are open to the idea?
You can work on implementation, but be prepared that it might be
much more difficult than you imagine. According to my experience
only very few developers managed to push their changes to the
review process.
Is there any way to provide our own implementations when they are missing in our code?
Yes. If it's a whole class, then simply take a look at `classlib` module and follow same naming pattern. Simply put missing class in your own codebase and this should work. For individual method of a class there's no solution.
And having own implementation of a class can make things a bit easier: for example, you not necessarily need general-purpose implementation of `CopyOnWriteArrayList` with all the corner cases properly covered. In this case you may write a (wrong) implementation that simply satisfies you. Of course, in the latter case I won't take such code in TeaVM code base.
Another way for you it to motivate me to implement these classes.
For example, you can convince me that helping you can make a lot
of buzz around TeaVM. Or you could become a sponsor.
Nacho Cordón
▶️ By saying "we" whom do you mean? Can you introduce yourself?
I'm a software engineer at Neo4j. We are working on a language server in Typescript, and we are transpiling (it's all experimental at the moment) part of the parsing stack of the database (JVM based) to provide for semantic analysis of queries. It was working fine with an old parser technology (which I suppose it was using very vanilla like Java), the new version (based on antlr4) is breaking because it's missing some of the implementations I listed.
▶️ Yes. If it's a whole class, then simply take a look at `classlib` module and follow same naming pattern. Simply put missing class in your own codebase and this should work. For individual method of a class there's no solution.
I think I found a way to circumvent that, even for missing methods. Creating a project (in our case maven project) that as you said overrides the classes we are missing but also just duplicates the code of the ones that are missing the methods (adding those missing methods) and importing that project instead of teavm-classlib seems to at least making it compile. But of course I'm cutting corners because we don't need a full functioning `CopyOnWriteArrayList` implementation for example.
▶️ Another way for you it to motivate me to implement these classes. For example, you can convince me that helping you can make a lot of buzz around TeaVM. Or you could become a sponsor
I had been chasing setting up a small sponsorship and I got that approved on Friday, but I understand it's not a big allocation so I would not expect that to be enough of a motivation. It's just to show we are really thankful for your project. Hopefully I can contribute some of those to the codebase myself. And if our project was successful even advertise how awesome TeaVM is!!
Alexey Andreev
▶️ By saying "we" whom do you mean? Can you introduce yourself?
I'm a software engineer at Neo4j. We are working on a language server in Typescript, and we are transpiling (it's all experimental at the moment) part of the parsing stack of the database (JVM based) to provide for semantic analysis of queries. It was working fine with an old parser technology (which I suppose it was using very vanilla like Java), the new version (based on antlr4) is breaking because it's missing some of the implementations I listed.
▶️ Yes. If it's a whole class, then simply take a look at `classlib` module and follow same naming pattern. Simply put missing class in your own codebase and this should work. For individual method of a class there's no solution.
I think I found a way to circumvent that, even for missing methods. Creating a project (in our case maven project) that as you said overrides the classes we are missing but also just duplicates the code of the ones that are missing the methods (adding those missing methods) and importing that project instead of teavm-classlib seems to at least making it compile. But of course I'm cutting corners because we don't need a full functioning `CopyOnWriteArrayList` implementation for example.
▶️ Another way for you it to motivate me to implement these classes. For example, you can convince me that helping you can make a lot of buzz around TeaVM. Or you could become a sponsor
I had been chasing setting up a small sponsorship and I got that approved on Friday, but I understand it's not a big allocation so I would not expect that to be enough of a motivation. It's just to show we are really thankful for your project. Hopefully I can contribute some of those to the codebase myself. And if our project was successful even advertise how awesome TeaVM is!!
I saw this, and now I have explanation why neo4j became a sponsor. Thanks! Sure, $25 is not enough to cover my work on these classes. However, this and the fact that TeaVM gets some more publicity is a good enough reason.