6502395: Is not a bug

[Emil Siemes]

The bug report 6502395 says: "Filer.getResource() states thatIOException is thrown if the file cannot be opened."
But
the attached Java class (see below) calls Filer.createResource() instead of
getResource() and when openReader() gets called an
IllegalStateException is thrown (Reading from a write-only Object). 
This
is actually expected behavior and not a bug .. JavacFiler.createResource
returns a FileObject. The spec for FileObject..createResource reads: 
"Throws:

IllegalStateException - if this file object was opened for writing and does not support reading 
UnsupportedOperationException - if this kind of file object does not support character access 
IOException - if an I/O error occurred"
The Object implementing FileObject returned by the JavacFiler is a com.sun.tools.javac.processing..JavacFiler$FilerOutputFileObject  and the openReader method clearly throws a IllegalStateException.So I think 6502395 should be closed and the bug effectively is in the bug report not in the jdk :-)

If
we change the attached Java class and replace createResource with
getResource and run it again with jdk7 (b36&b40 & b41) I get:
"java.io.FileNotFoundException: foo/SOURCE_OUTPUT.unexisting (No such file or directory)
    at java.io.FileInputStream.open(Native Method)
    at java.io.FileInputStream.<init>(FileInputStream.java:137)
    at com.sun.tools.javac.file.RegularFileObject.openInputStream(RegularFileObject.java:68)
    at com.sun.tools.javac.file.BaseFileObject.openReader(BaseFileObject.java:76)
    at javax.tools.ForwardingFileObject.openReader(ForwardingFileObject.java:92)
    at UnexistingResourceProcessor.process(UnexistingResourceProcessor.java"
as it should be.
With jdk6 update 11 I get instead:
"java.lang.UnsupportedOperationException
    at com.sun.tools.javac.util.DefaultFileManager$RegularFileObject.openReader(DefaultFileManager.java:1281)
    at javax.tools.ForwardingFileObject.openReader(ForwardingFileObject.java:74)
    at UnexistingResourceProcessor.process(UnexistingResourceProcessor.java"
Not
nice but may be still in line with spec because one could argue that a
FileObject for a non existing File does indeed not support character
access and therefore UnsupportedOperationException should be thrown.


Is wasn't able to find jdk releases older than build 36. Are they available for download somewhere? If yes I could check in which build the thrown exception changed. In b03 the DefaultFileManager was renamed to JavacFileManager maybe also the exception was changed around that time?

"Bug"-Description:

The spec for javax.annotation.processing.Filer.getResource() states that
IOException is thrown if the file cannot be opened.

So when we are trying to open nonexistent resource we should expect IOException thrown.

However getting nonexistent resource doesn't cause to IOException. Instead IllegalStateExceptionis thrown.

You may reproduce the behavior using following processor class:

package tmp;

import java.io.InputStream;
import java.io.Reader;
import javax.annotation.processing.AbstractProcessor;
import javax.annotation.processing.RoundEnvironment;
import javax.annotation.processing.Filer;
import javax.lang.model.element.TypeElement;
import javax.lang.model.SourceVersion;
import java.util.Set;
import java.util.HashSet;
import java.io.IOException;
import javax.tools.StandardLocation;

public class UnexistingResourceProcessor extends AbstractProcessor {
    StandardLocation[] locations = {
         StandardLocation.CLASS_OUTPUT, StandardLocation.SOURCE_OUTPUT
    };

    public Set<String> getSupportedAnnotationTypes() {
        return new HashSet<String>() {{add("*");}};
    }

    public SourceVersion getSupportedSourceVersion() {
        return SourceVersion.RELEASE_6;
    }

    public boolean process(Set<? extends TypeElement> annotations, RoundEnvironment roundEnv) {
        boolean passed = true;
        if(roundEnv.processingOver()) {
            Filer filer = processingEnv.getFiler();
            for (StandardLocation location: locations) {
                try {
// Opening for writing and then reading? No No No.
                    Reader reader = filer.createResource(
                            location, "", "foo/" + location + ".unexisting").openReader(true);
                    reader.close();
                    System.out.println("location: " + location);
                } catch (IOException expected) {
                } catch (Exception e) {
                    passed = false;
                    System.out.println("location: " + location + ", thrown: " + e);
                }
            }
        }
        return true;
    }
}








________________________________
Von: Jonathan Gibbons <Jonathan.Gibbons at Sun.COM>
An: Emil Siemes <digitalemil at yahoo.de>
CC: compiler-dev at openjdk.java.net
Gesendet: Mittwoch, den 10. Dezember 2008, 17:17:54 Uhr
Betreff: Re: 651779, 6502392 & 6502395: Annotation processing bugs

Emil,

Thank you for your offer; I will check up on your SCA for you.

The first bug number you gave is invalid; I'm guessing you mean
6517779.   I don't believe anyone is working on any of those issues.

-- Jon


On Dec 10, 2008, at 1:25 AM, Emil Siemes wrote:

Hi everyone,

I thought it might be more important than ever to start
collaborating and supporting openjdk now. 

I would like to volunteer for fixing annotation processing
bugs like 651779, 6502392 & 6502395. Anyone already working
on them?

I sent in a signed SCA by mail 10 days ago but haven't yet got any
feedback. Is that ok, or might the mail be lost? 

Thanks and best Regards
  Emil


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/compiler-dev/attachments/20081212/3fadb6d7/attachment.html