Subject: RE: [xsl] Saxon for windows? From: "Michael Kay" <mike@xxxxxxxxxxxx> Date: Sat, 4 Jun 2005 18:17:49 +0100 |
> > More than inefficient, it would be dangerous to run an XSLT > processor as > part of a *generally available* web service (I can only smile > envisioning > the kinds of creative DOS attacks that would be possible). > > Please explain how that would happen when the input is being > validated? Many people overlook the issue of extension functions. At one stage the W3C was running an XSLT transformation service on its web site where you could submit an arbitrary XSLT stylesheet and source document to be transformed. It was easy to establish (using system-property) that it was running xt, and was then easy to write a stylesheet that gave me a complete directory listing of the W3C web site by calling methods in the standard Java library. I could then have gone on to modify any file that the servlet code had access to. Saxon, and I imagine other good XSLT processors, has an option to disable extension functions so that it's safe to run untrusted stylesheet code. However, I think many people overlook this potential security weakness. And of course, there are cases where you need extension functions, which means you have to make sure the stylesheet code is trusted. Michael Kay http://www.saxonica.com/
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
Re: [xsl] Saxon for windows?, Dimitre Novatchev | Thread | Re: [xsl] Saxon for windows?, M. David Peterson |
[xsl] passing parameters with javas, Edwin Nys | Date | RE: [xsl] spacing issue, Michael Kay |
Month |