On Tuesday, 28. January 2003 18:58, S Woodside wrote:

> > "insert-after,/email[last()]"... but this is dangerous: you need some
> > way to
> > make sure you don't blindly execute xupdate statements someone nasty
> > might
> > craft. Even if you validate the resulting XML, the xupdate might DoS
> > your
> > server by executing very complicated statements. So best would be to
> > check
[...]

> When you say it's dangerous, I'm not sure how much ... people will
> never have access to documents other than their own docu, so they can't
> really trash anything except their own stuff. conditional processing is
> undefined in the WD, so how complicated can the xupdate get? I mean, I
> can check the length and make sure it's some sane size, but one of the
> huge advantages I want from this is to pass everything through the
> client, so that i can let xml "do the work" and not be hyper about
> verifying things.

Not verifying is calling for trouble - somewhere, somewhen. "Never trust user 
input" is true here as it is anywhere else.
Remember that ftp DoS bug? An "ls */../*/../*/../*/.." (continued some more) 
brought several ftpd's down to a halt by using 100% CPU. Are you sure that 
your XUpdate inplementation handles select="//*/..//*/..//*" efficiently? 
There could always be more hidden traps.

-- 
CU
  Joerg

PGP Public Key at http://ich.bin.kein.hoschi.de/~trouble/public_key.asc
PGP Key fingerprint = D34F 57C4 99D8 8F16 E16E  7779 CDDC 41A4 4C48 6F94


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to