I have GWT project that uses Generators to create light dynamic reflection objects.
I was wondering if anybody knows of a way to determine whether or not a particular class is referenced in the dependency tree beginning at all EntryPoints. If I could do this, I could avoid generating reflection data for classes that will never be used anyway.
My understanding is that when GWT does its compiling, it performs a similar check so that it can reduce the total size of the compiled code, but I haven't been able to find any related methods in TypeOracle or anything like that.
This is an indirect method of accomplishing what you are getting at. I believe each GWT module, is fully packaged into a regular java package. You can use
TypeOracle.findPackage(String pkgName)
to get the JPackage instance, and on that instance you use findType(String typeName) to see if a type is present in that package. If present, its likely that it is referenced in some file and GWT will compile it.
There is also this method getPackages() which returns an array of all packages known to this type oracle - therefore reachable for GWT compiler.
JPackage[] getPackages()
You can iteratively findType() on each package to find if the type is going to be compiled or not.
The BEST method is to define a custom annotation and whitelist all the classes that you do want to generate reflection code. You can annotate the required classes with it, and checking for that presence of annotation before generating code for it.
My favorite is to follow a naming convention over annotation, (I did both together), and thus maintain a whitelist, and make the convention (its usually a REGEX) a "setting" that can be changed however the team wants.
Related
I am using TreeHugger to generate code at runtime. I could not find many documents related to it. My question is, if I generate classes using treehugger, will I be able to access those classes in future?
To be precise: I want to read data coming from files like CSV and create classes at runtime . Can I use that class in future, say in the next class generated at runtime.
I am really new to scala, please forgive if I am not clear in explaining.
Thanks a lot!
I've done something similar, so I'll share what I've learned:
Treehugger ultimately generates code (strings) at runtime to be used in a subsequent, separate run (or I suppose to be eval'd at runtime, but I never got that to work).
So the course of action depends on what you mean by "runtime":
Are your .csv files only available at runtime? If you have access to the files at compile time (as is often the case), then are examples of your two options: experimental (scala macros) or traditional (sbt plugin) -- both approaches are similar but have subtle pros and cons.
If you only have access to the files at runtime, but still need to generate and "type" the classes and make the compiler expect them, then it seems to me that somebody has made a bad design mistake! But if you find yourself stuck in this circumstance, then it is possible to define and load classes at runtime with a bytecode-engineering library and some type-checker black magic (runtime type provider).
I'm building a tool that will receive unpredictable data structure, and I want to generate case class to accomplish the structure of the received data.
I'm trying to figure out if it's possible to generate case class at runtime? This structure will be know only at runtime.
It's something similar to what macro does, but in runtime.
I've found this project on the internet
mars
Which is very close to what I want to do ,but I couldn't find if it was successful of not.
Another way of doing it is generate the code, compile and put the result in the classpath, like IScala is doing to use the code in an iterative way. But I don't think that this will scale.
Does anybody has already done something like runtime code generation?
This question was also posted in scala-user mailing list
UPDATE: (as per the comments)
If all you want is throw-away code generated at runtime to be fed into to a library that cannot work with just lists and maps, and not code to be stored and used later, it would make sense to look for solutions to this problem for Java or JVM. That is, unless the library requires some Scala specific features not available to vanilla JVM bytecode (Scala adds some extras to the bytecode, which Java code doesn't need/have).
what is the benefit of generating statically typed code dynamically? as opposed to using a dynamic data structure.
I would not attempt that at all. Just use a structure such as nested lists and maps.
Runtime code generation is one of the purposes of the Mars Project. Mars is under development, at the moment there is no release version. Mars requires its own toolchain to expand macros at runtime and should use several features unique to scala.meta (http://scalameta.org/), for example, AST interpretation and AST persistence. Currently we are working on ASTs typechecking in scala-reflect, required for runtime macros expansion.
I have read that for GWT, specifying methods to return a concrete implementation, for example:
public ArrayList<String> getList();
instead of the normally-preferred "abstract interface", for example:
public List<String> getList();
results in GWT producing a smaller compiled javascript file, because the client (ie js) code doesn't have to cater for all known implementations of the interface (in the example of List, the client code would have to be able to handle LinkedList, ArrayList, Vector, etc), so it can optimize the js by not compiling unused implementations.
My closely-related questions are:
Is this true? (the following questions assume it is true)
Is the optimization per-class that uses interfaces, or per application? ie
Do I see a benefit just refactoring up one class? or
Do I only see a benefit once all client classes are refactored to not use interfaces?
The following assumes that you use the interface as part of the signature of GWT RPC service. I think if you do not use the interface in the signature of GWT RPC service, the effect of using classes instead of interfaces should be minimal (e.g. the GWT compiler will only compile the used implementations)
Is this true? (the following questions assume it is true)
Yes, the output of the GWT compiler gets smaller when it 'knows' better which classes might be send from server to client.
Is the optimization per-class that uses interfaces, or per application? ie
In case of GWT RPC, per application.
Do I see a benefit just refactoring up one class?
Yes, one interface replaced by an implementation can reduce generated code size by a few kb, if the interface would require to include code for many classes.
However, apart from using implementations instead of interfaces, also a 'blacklist' of classes can be defined in the module definition file to explicitly circumvent the inclusion of implementations in the generated code: something like
<extend-configuration-property name="rpc.blacklist"
value="-java.util.ArrayList" />
I just did a test based on the sample app generated by webAppCreator, but I added 3 simple services that returned either List<String> or ArrayList<String>, depending on the build.
The results were that having all services use ArrayList<String> saved about 5Kb from the compiled javascript over having any mix of the return types.
That proves the saving is real and per-app (not per-service).
It also show how much it saves (in this case).
This doesn't actual to the GWT-compiler in general. Such approach is applied only for classes used with code generation. For example, when using Remote Procedure Calls. See this question for more detail information. Thus, if you declare an interface instead of a concrete class as the return type, the compiler includes all possible implementations in your compiled code. This increases time of compilation and a amount of generated code.
Actually one might develop application using GWT without RPC. In this case compiled code doesn't bloat when using interfaces.
I have just finished the first version of a Java 6 compiler plugin, that automatically generates wrappers (proxy, adapter, delegate, call it what you like) based on an annotation.
Since I am doing mixed Java/Scala projects, I would like to be able to use the same annotation inside my Scala code, and get the same generated code (except of course in Scala). That basically means starting from scratch.
What I would like to do, and for which I haven't found an example yet, is how do I generate the code inside a Scala compiler plugin in the same way as in the Java compiler plugin. That is, I match/find where my annotation is used, get the AST for the annotated interface, and then ask the API to give me a Stream/Writer in which I output the generated Scala source code, using String manipulation.
That last part is what I could not find. So how do I tell the API to create a new Scala source file, and give me a Stream/Writer/File/Handle, so I can just write in it, and when I'm done, the Scala compiler compiles it, within the same run in which the plugin was invoked?
Why would I want to do that? Firstly, because than both plugins have the same structure, so maintenance is easy. Secondly, I want to open source it, and there is just no way to support every option that anyone would want, so I expect potential users to want to extend the generation with their own code. This will be a lot easier for them if they just have to do some printf(), instead of learning the AST API (this also applies to me).
Short answer:
It can't be done
Long answer:
You could conceivably generate your source file and push that through a parser instance within your plugin. But not in any way that's likely to be of any use to you, because you'd now have a bigger problem to contend with:
In order to grab all the type/name information for generating the delagate/proxy, you'll have to pick up the annotated type's AST after it has run through both the namer and typer phases (which are inseperable). The catch is that any attempts to call your generated code will already have failed typechecking, the compiler will have thrown an error, and any further bets are off.
Method synthesis is possible in limited cases, so long as you can somehow fool the typechecker for just long enough to get your code generated, which is the trick I pulled with my Autoproxy 'lite' plugin. Even then, you're far better off working with TreeDSL to generate code instead of pumping out raw source.
Kevin is entirely correct, but just for completeness it's worth mentioning that there is another alternative - write a compiler plugin that generates source. This is the approach that I've adopted in Borachio. It's not a very satisfactory solution, but it can be made to work.
Edit - I just reread your question and realised that you're actually asking about generating source anyway
So there is no support for this directly, but it's basically just a question of opening a file and writing the relevant "print" statements. There's no way to invoke the compiler "inside" a plugin AFAIK, but I've written an sbt plugin which hides most of the complexity of invoking the compiler twice.
Lets say I have a class...
com.mycom.app.AbstractMessage
There is another class in
com.mycom.model.QueryResponse
QueryResponse extends AbstractMessage and notice they are in different pacakges
com.mycom.model is a GWT Module and in the module XML
When I compile model there are errors. However when I try to use QueryReponse in another GWT module, I get runtime errors
"No source code is available for type com.mycom.app.AbstractMessage; did you forget to inherit a required module"
This lends me to believe that AbstractMessage was not compiled/compiled right to begin understandably because I DO NOT WANT to have "app" package be a GWT module
In other words, I only want to compile all classes in "model" and not any super classes. How can I tell the GWT compiler/rpc/linker/serializer etc not to do so?
i.e Is there a way to tell GWT not to walk beyond certain classes when it serializing/compiling it
I am doing this a source environment where we have a lot of packages, most of them depend on MODEL only and I DO NOT want to make a GWT module out of every package, just so it compiles.
Thoughts anyone?
I did a little bit of research on this one, you are right GWT will look for all implementations of an Abstract class, if and only if, the AbstractClass is referenced in an RPC GWTAsync interface, even though some are in non-GWT packages.
Let's say an object of type AbstractClass comes in over the network, and the GWT deserializer is now tasked with coverting the network data into a specific instance. It needs to know about all implementations of AbstractClass, to find which is coming over the network right now! -- So to accomplish this it, at compiletime, generates a .rpc file for each GWT service interface, listing all possible concrete types that the service methods can return.
Ray Ryan (Google employee) once mentioned that it is a bad idea to use interfaces arguments or return types in any RPC interface. - because it makes it difficult for the deserializer to know the exact type.
You can hand edit the generated RPC file and remove the offending types, or mark the other implementations as Non Serializable by not implementing Serializable in those implementations in other packages.
A Better way could be - I suspect you wrote code : "implements java.io.Serializable" at the top level (for the AbstractClass itself), maybe it's now time to move it to each implementation.
Now the GWT RPC deserializer's task is clear and straightforward - it knows that only certain implementations (that are serializable) of the AbstractClass will come over the network, and reach and compile them only. So it will not compile the other non serializable subclassess of your AbstractClass - as it knows they arent serializable.
There is one more option : If as I suspect you are using the command pattern - I have seen all the abstract interfaces, super classes for Command and Response etc always go in the client side packages - i.e., those that are GWT compiled. They are referrable and usable and instantiable for the server end of the application - so these source files are compiled twice, once by GWT into javascript for browser usage, and once by javac into bytecode for allowing reference from serverside. Thus in all GWT modules, including gwt-user.jar if you open them with 7Zip or WinZip you will see source and class files JARed together.
I recommend Moving AbstractMessage into the models package - as it is the model QueryResponse's super class.
And also inhertance in models is only a good idea, if you have 0 fields and only methods(behaviour) in the super class.
Lastly, if GWT is to make your QueryResponse into javascript - it needs ALL Types mentioned in the source file, to compile properly. So do not mention any server-only-classes in a source file meant to become javascript.
Have a region that has all the server-side java classes that will be run in a JVM on the server, and another region full of source files that will be compiled into javascript by the GWT compiler. The server-side region code/classes CAN refer to client region code/classes but defenitely NOT the vice versa. Make sure that no code thats gonna become javascript is referring (even an unused import statement) to a server side class.
GWT compiler works with source files only, however you need to compile client code into .class files so your serverside classes can refer to them.
NEW EDIT :
I did a little bit of research on this one, you are right GWT will look for all implementations of an Abstract class, if and only if, the AbstractClass is referenced in an RPC GWTAsync interface, even though some are in non-GWT packages.
Let's say an object of type AbstractClass comes in over the network, and the GWT deserializer is now tasked with coverting the network data into a specific instance. It needs to know about all implementations of AbstractClass, to find which is coming over the network right now! -- So to accomplish this it, at compiletime, generates a .rpc file for each GWT service interface, listing all possible concrete types that the service methods can return.
Ray Ryan (Google employee) once mentioned that it is a bad idea to use interfaces arguments or return types in any RPC interface. - because it makes it difficult for the deserializer to know the exact type.
You can hand edit the generated RPC file and remove the offending types, or mark the other implementations as Non Serializable by not implementing Serializable in those implementations in other packages.
A Better way could be -
I suspect you wrote code : "implements java.io.Serializable" at the top level (for the AbstractClass itself), maybe it's now time to move it to each implementation.
Now the GWT RPC deserializer's task is clear and straightforward - it knows that only certain implementations (that are serializable) of the AbstractClass will come over the network, and reach and compile them only. So it will not compile the other non serializable subclassess of your AbstractClass - as it knows they arent serializable.