The bug for this issue was accepted: JDK-8255918. I have made the change that I suggested in my original email and tested, but have not yet opened a pull request. Please let me know what (if anything) needs to occur for the CSR process due to the method signature change (added `throws`).
Thank you, Mike On Wed, Oct 28, 2020 at 12:52 PM Joe Wang <huizhe.w...@oracle.com> wrote: > Hi Mike, > > As you said, creating a bug report would be a good start. If it involves > a signature change, it'd need to go through a proper review (CSR) process. > > When you are ready to submit a bug report, please make sure to add a > test case to illustrate the use case scenario. > > Thanks, > Joe > > On 10/28/20 5:14 AM, Michael Edgar wrote: > > Hi everyone, > > I'm working on a project that makes use of the StAX API and an issue I > have > > encountered is that when wrapping an `XMLStreamReader` with a > > `StreamFilter`, errors encountered in the setup are not thrown to the > > caller. The source of the error could be any stream error that is > triggered > > as the `XMLStreamFilterImpl` advances to the next acceptable event. > > Ultimately, when attempting to utilize the filtered reader, some > secondary > > exception will occur, but the original `Exception` is lost. > > > > I have not seen any other issues related specifically to this problem, > so I > > would like to propose removal of the try/catch in the constructor of > > `com.sun.org.apache.xerces.internal.impl.XMLStreamFilterImpl` and the > > method signature changed to declare that `XMLStreamException` is thrown. > > The constructor is used by > > > `com.sun.xml.internal.stream.XMLInputFactoryImpl.createFilteredReader(XMLStreamReader, > > StreamFilter)` which itself already declares the same exception and is an > > implementation of the public `XMLInputFactory` interface. > > > > Further, the `nextTag` method of the same class has a bug where it checks > > for `START_ELEMENT` events twice. > > > > I have an OCA in place and I am happy to submit a PR, but I believe that > a > > bug record needs to be opened in order to proceed. > > > > Thank you, > > Mike > >