RE: Regarding dynamic execution of expressions & security (Was: Re: [F&O] Casting to xs:NOTATION, xs:QName not clarified)

It is general practice (and a requirement in my opinion), that fn:doc()/fn:document() should use a URI resolver that provides enough security mechanisms to allow programmers to deal with this issue.

Best regards
Michael

> -----Original Message-----
> From: Laurens Holst [mailto:lholst@students.cs.uu.nl]
> Sent: Thursday, February 03, 2005 12:30 AM
> To: Michael Rys
> Cc: public-qt-comments@w3.org
> Subject: Regarding dynamic execution of expressions & security (Was: Re:
> [F&O] Casting to xs:NOTATION, xs:QName not clarified)
> 
> Michael Rys wrote:
> > The problem is that in some implementation environments, dynamic
> > execution of expressions is considered a security risk and it is not
> > clear how this will relate to static typing of the query and some other
> > issues. The WG has decided to not standardize this aspect in this
> > version to gain more experience with the existing language feature and
> > to maybe adding it at a later point (vNext).
> 
> A small comment regarding the security risk argument: XSLT allows access
> to external documents using the document() function. These document URIs
> are regular strings, which can be taken from the document (and
> frequently are, e.g. when rendering multiple documents based on an XML
> file with a TOC), and are not necessarily limited to local paths. This
> basically allows access to arbitrary external documents and IMHO this is
> a much larger security risk, yet that didn't prevent standardisation.
> 
> My 2¢.
> 
> 
> ~Grauw
> 
> --
> Ushiko-san! Kimi wa doushite, Ushiko-san nan da!!

Received on Thursday, 3 February 2005 19:29:43 UTC