[quagga-dev 4162] Re: example libzebra zclient utility

Andrew J. Schorr aschorr at telemetry-investments.com
Thu Jun 8 21:10:30 BST 2006

Hi Paul,

On Tue, May 30, 2006 at 07:40:04PM +0100, Paul Jakma wrote:
> Hope you enjoyed your hols ;)

Yes, I did, thank you.  Sorry to be so slow to respond.

> I tinkered a bit with the redistribute strings thing, to add a 
> 'route_types.txt' table, able to hold all the required information 
> (yat another gawk script ;) ):
>  http://cvs.quagga.net/cgi-bin/gitweb.cgi?p=people/paul/quagga;a=commitdiff;h=d4b0f18f9c96177e5a255d17f36f6e945820cd43;hp=0e446afce8b0e4d9de0f0e5714d3f73944c09de3

Yes, that basically looks good, although I haven't tried it or checked
the code.  But I'm asssuming it does what it's supposed to do.

> Which allows client daemons to be excised of redist strings, e.g. 
> ospfd:
>  http://cvs.quagga.net/cgi-bin/gitweb.cgi?p=people/paul/quagga;a=commitdiff;h=cdddef88ac00e7c46b2e02ff4b4309e0f2a3b0ad;hp=d4b0f18f9c96177e5a255d17f36f6e945820cd43

That looks great to me.

> It only generates defines for the redistribute strings at the moment, 
> tailored to each daemon according to the table, but if that was way 
> forward, it ought to also generate the table log.c uses.


> Also, all redist strings get '<1-255>', so that in the worst case 
> users at least can specify the ZEBRA_ROUTE_* number directly. (We'd 
> want to add a command to show the known daemon<->number numbers 
> probably).

I'm not sure whether this number feature is needed, but I suppose it
can't hurt.  But I wouldn't invest too much effort in it.  Is there
some special reason you want that?  Just out of fear that the logic
to build the redist strings may not work correctly?

> Right or wrong way forward? :)

Yes, I think this is the right way to go (a table declaring the attributes
of each protocol, and then some awk code to generate the endlessly repeated,
mindlessly repetitive strings based on that).

Great work!


More information about the Quagga-dev mailing list