Thomas, Not sure if there's a better way to fix XERCESJ-809, but agree that what was done is trading one problem for another.
If you're stuck, you could always build Xerces from source, reversing the patch for XERCESJ-809. Thanks. Michael Glavassevich XML Parser Development IBM Toronto Lab E-mail: mrgla...@ca.ibm.com E-mail: mrgla...@apache.org Thomas Scheffler <thomas.scheff...@uni-jena.de> wrote on 07/20/2011 08:30:28 AM: > Am 20.07.2011 13:26, schrieb Michael Glavassevich: > > > > Thomas, > > > > Have you read this thread [1]? What you're observing for > > EntityResolver2 is probably due to the same change. > > Hi, > > yes this is exactly the same issue I ran to it here. Don't know if it is > a good idea to change the behavior here. I use the EntityResolver2 > interface as this allows systemId to be relative and be interpreted by > the implementation as opposed to the EntityResolver interface, where it > is always expanded to an absolute URI. If xerces as no clue about the > baseURI (source is inputstream, e.g. blob from database), it expands it > to the current directory, where the Java process is started. Or else it > uses the baseURI to expand it by itself. > For example the standard parser of Oracle Java 6 SDK, does not do that. > It is extremely sad, that Xerces 2.11.0 puts some logic into it, when > this is expected to handled by the implementation of EntityResolver2. > What makes this even more inconsistent is the fact, that XML Schema > documents requested by other XML Schema documents are indeed requested > as a relative URI. > Until 2.11.0 I could trust that a absolute URI is valid and that > relative URIs could be loaded as a resource from the classpath of my > implementation. Now I have to restore the "relative information" by > guessing it from the supplied URI. And guessing is most times a bad thing. > > I wonder if there is no other way to fix XERCESJ-809, without breaking > compatibility. Apart of that I could not understand the description of > the mentioned bug. > > regards > > Thomas Scheffler > > > [1] http://xerces.markmail.org/thread/zpazk5bukquyijan > > > Hi, > > > > > > after an upgrade from xerces 2.9 to 2.11.0 I noticed a different > > > behavior with a EntityResolver2 implementation. > > > > > > If I parse an XML file with a schema defined with a relative > > > systemId, Xerces always resolves the systemId to an absolute one. > > > This is the behavior of the EntityResolver interface and should not > > > be the same with EntityResolver2, where the implementation should > > > resolve relative systemIds. > > > > > > Should I file a bug for this? This is a rather urgent problem and > > > I cannot move back to 2.9.0, as this version cannot validate MODS > > > documents correctly. > > > > > > regards > > > > > > Thomas Scheffler > > > > > > --------------------------------------------------------------------- > > > > > > > To unsubscribe, e-mail: j-users-unsubscr...@xerces.apache.org > > > For additional commands, e-mail: j-users-h...@xerces.apache.org > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: j-users-unsubscr...@xerces.apache.org > For additional commands, e-mail: j-users-h...@xerces.apache.org
