[frr] The bgpafisafi pull request
philippe.guibert at 6wind.com
Thu Jan 26 10:34:34 EST 2017
I understood that the way this function was used, was like the other vty
- get afi/safi values
- get afi/safi/vrf values.
meaning that the caller of the function must ensure that the values he
passes as params have good initial values.
I would have done a little change.
This is because you handle pointers in that function (*vrf), and you assume
that this pointer is valid. Whereas an user that does not want to care
about VRF, does not want to initialise a specific value. he uses NULL as
For that, I would have ignored vrf case ( returned 0 if a vrf is found, and
return ok if no vrf is found).
The rest of the series is ok for me.
On Thu, Jan 26, 2017 at 3:21 PM, Donald Sharp <sharpd at cumulusnetworks.com>
> Philippe -
> You had a review comment two days ago:
> how do you initialise vrf in the case vrf_name is already null (
> commands with no vrf at all)?
> I would lie the *vrf value to be set inside that function.
> I responded with:
> I do not believe this is the right thing to do. Only the calling
> function knows what the right defaults are for the afi/safi/vrf due to
> we assume different defaults based upon the command. This approach
> would break the ability for the calling function to do the right
> thing. As I stated I will update the Pull Request with better
> documentation for this funciton.
> I've also added a commit that updates the function to further explain
> it. Is there anything else you would like to have done?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the dev