Craig, yes, that's exactly the problem. Valve is another prominent case where the attribute-checking is not possible.
One solution, but I confess that I would not recommend it, is to distinguish between the different types, i.e. change <Valve> to <AccessLogValve>,<RequestDumperValve>,<RemoteHostFilter> etc. That would certainly make the server.xml validatable, but create the burden of changing the xsd/dtd every times a user creates her own Valve/Logger/Realm etc. Could xslt be a solution to check the required attributes if the dtd/schema uses union? Maybe that is to much effort because anyway if a required attribute is not present, the digester would moan. Mika ----- Original Message ----- From: "Craig R. McClanahan" <[EMAIL PROTECTED]> To: "Tomcat Developers List" <[EMAIL PROTECTED]> Sent: Friday, November 30, 2001 4:09 AM Subject: Re: server.xml DTD/Schema > One thing to remember is that it is not technically possible to write a > DTD for server.xml that covers all possible cases (and I suspect that's > true for Schema as well). Consider the following cases: > > * Elements like <Logger> and <Realm> let you define which implementation > class you want, from the set of choices included with Tomcat. The set > of attributes that are valid depends on which implementation class you > choose -- and there is no way to make that distinction in a DTD. The > best you could do is list the union of all possible attributes -- but > that is not semantically valid for any single implementation. > > * Even more generally, Tomcat users are free to install their own > implementations of Tomcat classes, and there's no way your general > purpose DTD would know which attributes are valid. > > Craig McClanahan > > > > On Fri, 30 Nov 2001, Mika Goeckel wrote: > > > Date: Fri, 30 Nov 2001 01:01:46 +0100 > > From: Mika Goeckel <[EMAIL PROTECTED]> > > Reply-To: Tomcat Developers List <[EMAIL PROTECTED]> > > To: [EMAIL PROTECTED] > > Subject: server.xml DTD/Schema > > > > Hi, > > > > I've built a first version of a DTD/Schema for server.xml and would ask if > > someone would like to review it? > > > > I would prefer the Schema, because it allows more checking, but I haven't > > seen a parser which checks against schemes, so I created a DTD from it as > > well. > > > > As this is quite a bunch of lines, please hands up who wants to receive it. > > > > Cheers, Mika > > > > P.S.: The initial cut is from the docu, I plan to go through the source > > tomorrow to recheck. > > > > > > -- > > To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> > > For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> > > > > > > > -- > To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> > For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>