Quoting "Craig R. McClanahan" <[EMAIL PROTECTED]>: > David Graham wrote: > > >--- Robert Leland <[EMAIL PROTECTED]> wrote: > > > > > >>Currently the Validator PlugIn doesnâEUR^(TM)t validate the XML file. I > have > >>updated it to remove the deprecated methods and to validate the XML > >>files. > >> > >>Here is the question: Currently, the validator plugin fails by logging a > >>message but doesnâEUR^(TM)t take advantage of throwing an exception. I > believe > >>it should throw an exception if the validator file is invalid. Are there > >>any thoughts or objections to throwing an exception, is that too > >>incompatable with its current behaviour ? > >> > >> > > > >I agree that we should throw the exception. IMO, it should fail fast and > >loud at startup if you've misconfigured the validations. > > > > > I agree with this as well, but of course the same concept should apply > to all our other configuration information. How many of us have > mistyped the name of a form bean class, for example (raises hand :-). > > There was an interesting thread on a similar situation with Hivemind on > COMMONS-DEV last week, where those developers seem to believe that users > shouldn't want such support ever. I don't agree with that extreme a > viewpoint, but it might be nice to make such checks a configuration > parameter option (I'd vote for default=true because it probably won't > hit startup performance enough to matter for most people). >
One issue with this is to make sure the validator DTD is up to date. The reason I say this is because the DTD appears to be missing the var-jstype sub-element of the var element. Even the 1.2.0 version of the DTD on CVS is missing this. I assume it is a valid element, because even though I can't validate my document, the element is used properly when I set it. Matt Bathje --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
