[Nsis-imp] Re: Question on QSPEC Implementation (was RE: [NSIS] NSIS for end users - It's reality)

[Bernd Schloer <[e-mail]>]

Hi Jerry and all,

:-)

I believe that this is not a big problem. Even if the RMF parses the
parameters sequentially, the whole object can be accessed any time. You
can do the manipulation on the object level or you can consider the
QSPEC as a (char-)array and the manipulation can be done on the byte-
or bit-level. And it is passed as an entire object back to the NSLP.

Bernd


Ash, Gerald R (Jerry), ALABS wrote:
> Christian, All,
>  
> Congratulations on this significant accomplishment!
>  
> We have a question regarding QSPEC implementation.  In Section 4.3.3.1 (Reservation Failure & Error E-Flag) of the QSPEC draft http://ietf.org/internet-drafts/draft-ietf-nsis-qspec-11.txt <http://ietf.org/internet-drafts/draft-ietf-nsis-qspec-11.txt>  we describe the dual use of the E-flag.  
> 
> We'd like your opinion and others' opinion on the NSIS implementation list RE a possible issue with the 'overloading' of the E-Flag.  As described in Section 4.3.3.1 the E-flag has a dual purpose and can mean either 'QSPEC parameter not met' or 'QSPEC parameter error'.  As described, if the E flag is set for erroneous parameters, it MUST NOT be set for parameters not met, and visa versa.  I.e., the use of the E flag is completely disjoint for the two cases.  
> 
>  
> 
> However, what if a QSPEC contains n parameters. The RMF parses the parameters sequentially, and the first parameter is not met and its E-Flag is raised.  The parsing continues, everything is fine.  But the nth parameter is erroneous. Its E-Flag is raised.  What we currently require is the RMF goes back to the first parameter and unsets the E-Flag.
> 
>  
> 
> Is this an implementation problem big enough that it would warrant an additional flag to distinguish the 2 cases?  We would prefer not to introduce another flag unless this is seen as a significant problem.
> 
>  
> 
> Your opinion on this would be helpful since you have already developed a prototype QSPEC implementation.
> 
>  
> 
> Thanks,
> 
> Jerry, Cornelia, Attila
> 
>  
> ________________________________
> 
> From: Christian Dickmann [mailto:[e-mail]] 
> Sent: Thursday, September 28, 2006 7:15 AM
> To: [e-mail]; [e-mail]
> Cc: [e-mail]
> Subject: [NSIS] NSIS for end users - It's reality
> 
> 
> 
> Hi,
> 
>  
> 
> the NSIS team of the University of Goettingen is proud to announce the release of a new development version of our NSIS package, which offers a lot of new features. We now support Traffic Control (tc) on Linux to really reserve resources and allow the integration of NSIS/QoS into applications by offering a QoS NSLP API. This development version also comes with a lot of bugfixes and updates to new draft versions of GIST and QoS NSLP. 
> 
>  
> 
> However, the most exciting new feature is the support for Windows XP and OpenWRT[1]!
> 
>  
> 
> The Windows XP version is a demonstration prototype, which is able to detect the UDP port used by Skype[2] and trigger the local QoS NSLP daemon to request resources for phone calls.
> 
> In addition, we support OpenWRT, which is a Linux distribution for embedded devices, i.e. broadband Wireless LAN / DSL routers. On a lot of such routers, like the Linksys WRT54G (see the OpenWRT homepage[1] for a list of supported hardware), OpenWRT can be installed as an alternative firmware.
> 
>  
> 
> All new features combined, NSIS usage in SOHO networks becomes reality! 
> 
> The Windows machine triggers the request and the DSL router reserves the resources by the use of the Traffic Control tool.
> 
>  
> 
> You can find the development versions for all platforms of the NSIS package on our beta version page. The ChangeLog which lists all changes is also available on that page. Please see the individual README files on how to use the packages.
> 
>  
> 
> http://user.informatik.uni-goettingen.de/~nsis/beta.html
> 
>  
> 
> We encourage anyone to try this development version and appreciate feedback of any kind, such as installation/setup experience, bug reports, feature requests, etc.
> 
>  
> 
> Christian Dickmann
> 
> On behalf of University of Goettingen NSIS team
> 
>  
> 
> [1] OpenWRT: http://www.openwrt.org
> 
> [2] Skype: http://www.skype.com
> 
>  
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> nsis mailing list
> [e-mail]
> https://www1.ietf.org/mailman/listinfo/nsis

_______________________________________________
NSIS-imp mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/nsis-imp