On Fri, Sep 08, 2006 at 06:16:29PM +0200, Mathieu Roy wrote:
> Hello,
> 
> I'm deep in the code right now, as I'm at CERN.
> 
> First of all, thanks Sylvain and Tobias for the good work done.
> 
> The overall seems very good.
> 
> However, I'm puzzled to see some new $_COOKIE[] getting into the code, and a 
> few things like that. Please, read recipe #122
> 
>       https://gna.org/cookbook/?func=detailitem&item_id=122
> 
> For instance, it states 
> 
> "USER INPUT SANITIZATION
>  ---------------------------
>  It is important to always obtain user input from sane_*() functions.
>  
>  The goal is to get Savane working with register_globals set to off. Which 
> means all variables that can be provided as user input should not come as 
> global but always from sane_*() functions, like sane_post() or sane_cookie().
>  
>  Gradually pages will be register_globals off compliant. The ones that are 
> have a call to register_globals_off() after the beginning requires. It is 
> absolutely forbidden to remove such call. Pages that are made compliant with 
> register_globals set to off should stay that way."
> 
> It may seems to you the same to use $_COOKIE[] than sane_cookie(), but it is 
> not. Considering the lenght of the code, it's not for tomorrow that it will 
> be register globals off compliant. The way to proceed is to use the sane_*() 
> functions that will be register globals off  but that will also work with 
> register_globals_off(), unlike $_COOKIE[]. Each time you add a $_COOKIE[], 
> which would be fine if all the software was register globals off compliant, 
> you add one piece of code that will have to be replaced by sane_cookie().

If the processed data is not meant to contain quotes and is validated
before used, there will be no problem - but why not.

I see however that
http://fr.php.net/manual/fr/security.magicquotes.php is going to be
removed in PHP6, which means we better prepare to set it to false by
default in sample|debian configuration files.


> BTW, if you rewrite a lot some pages, do not hesitate to add the 
> register_globals_off() on top and benefit of the change to have one more page 
> compliant.

It seems that this function gives the user a way to kill any global
variable, I don't think that's good.

-- 
Sylvain

_______________________________________________
Savane-dev mailing list
[email protected]
https://mail.gna.org/listinfo/savane-dev

Reply via email to