On Wed, Mar 02, 2016 at 06:09:45PM +0000, Paul Jakma wrote: > On Wed, 2 Mar 2016, Paul Jakma wrote: > > > Someone mentioned Vyatta had done some work in this area. I think Cumulus > > might have done some work on other interfaces into things too, but I'm not > > sure. > > Oh, and David Lamparter was also tinkering on stuff. There was a > discussion recently.
Indeed, the work I had announced several months ago has come to bear some results; I need to send a writeup on that to the list. As for this discussion, the approach that seemed most fruitful in my view was to push transaction functionality outside of Quagga. There's no advantage to having it inside, since when it's properly modularised it looks like this: CLI ----> transaction code ----> settings API (or other config) And when the settings API is properly isolated, there's no point in having the transaction code in the same process (or codebase, or language) -- you can just put some nice IPC there. Any point where a transaction handler could be attached is also a point where IPC can be attached to modularize things. This is IMHO a very strong argument against adding much in this front. I believe the Cumulus approach makes the same argument, showing that this functionality doesn't need to be shoehorned in at brute force. -David _______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
