That is true. But it also means you aren't validating much. Ralph
On Sep 26, 2012, at 8:25 AM, Paul Benedict wrote: > IIRC, XSD has an <xsd:any> type that allows you to accept any element. This > allows a flexible schema. > > On Wed, Sep 26, 2012 at 10:04 AM, Ralph Goers <ralph.go...@dslextreme.com> > wrote: > Have you guys read > http://logging.apache.org/log4j/2.x/manual/configuration.html#ConfigurationSyntax? > In particular the "strict" and "schema" attributes. > > IMO, using a schema for the concise format would be very impractical. > Personally, I'm not sure it makes sense even for the strict format as it > would have to "know" about the attributes the various Appenders and Filters > can accept to be of any value. The schema would also have to be updated as > new components are added. > > Ralph > > > On Sep 26, 2012, at 6:55 AM, Paul Benedict wrote: > >> It's true that validation can add a bit of grogginess to the system. >> However, that can simply be controlled by a Java system property if >> necessary. I don't think we need the validation at runtime, however, having >> an XSD at design time is a definite. I need all the help I can get inside >> Eclipse :-) >> >> If not already, we should provide an XSD in the log4j api or impl jar. >> >> Paul >> >> On Wed, Sep 26, 2012 at 8:07 AM, Ivan Habunek <ivan.habu...@gmail.com> wrote: >> On 26 September 2012 13:39, Gary Gregory <garydgreg...@gmail.com> wrote: >> > Incorrect! >> >> I stand corrected. Still, it's probably not wise to have automatic XSD >> validation before configuration because of performance issues. But it >> would be nice to have a method which users can call and validate their >> XML configuration files manually. >> >> Regards, >> Ivan >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org >> For additional commands, e-mail: log4j-dev-h...@logging.apache.org >> >> > >
