Hi csit-dev and Paul, This Changes also caused CSIT test fails, Should it be fixed in vpp_api_test ?
[cid:image002.jpg@01D4F6C8.0F022D40] It can be seen that “vpp_sw_index : None” , but the sw_interface_set_flags try to set “None” link up. Best Regards Yulong Pei From: vpp-dev@lists.fd.io [mailto:vpp-dev@lists.fd.io] On Behalf Of Paul Vinciguerra Sent: Friday, April 19, 2019 5:51 AM To: Matthew Smith <mgsm...@netgate.com> Cc: vpp-dev <vpp-dev@lists.fd.io> Subject: Re: [vpp-dev] sw_interface_dump changes Hi Matt. If you look at https://gerrit.fd.io/r/#/c/18693/5/test/vpp_papi_provider.py, the field was set to default to ~0 for the python client. When I made the change, I chose to do it this way to make the api fall more in line with other calls such as l2_interface_efp_filter and sw_interface_rx_placement_dump. ~0 is the sentinel value typically used for "any" or "all" interfaces. This change was made more backward compatible in: https://gerrit.fd.io/r/#/c/18980/. Paul On Thu, Apr 18, 2019 at 3:22 PM Matthew Smith <mgsm...@netgate.com<mailto:mgsm...@netgate.com>> wrote: Hi, It looks like in https://gerrit.fd.io/r/#/c/18693/ the message format and handler for sw_interface_dump changed in a backwards-incompatible way. It's not too difficult to correct for, but it seems like any API clients that used this call might be broken. Some code that I maintain that connects to the binary API broke, and vpp_api_test looks like it did too. I tested with a build generated from commit 5a8844b "GRE: API update". A field was added to the message named sw_if_index, which causes a single interface to be dumped. You have to set this to ~0 to receive all interfaces, which was the default behavior before. Since clients written against earlier versions of the API didn't have to populate this field, they are probably passing 0 because the memory was zeroed when it was allocated. This has the effect of only receiving the details for sw_if_index 0 ("local0") when a sw_interface_dump message is sent in which the value of sw_if_index has not been set. I have adjusted my code to deal with the change but I don't know how many other clients of the binary API exist out there that will be broken by this. Should backwards compatibility be restored? Or just fix vpp_api_test and move on? -Matt -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#12817): https://lists.fd.io/g/vpp-dev/message/12817 Mute This Topic: https://lists.fd.io/mt/31234917/1594641 Group Owner: vpp-dev+ow...@lists.fd.io<mailto:vpp-dev%2bow...@lists.fd.io> Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [pvi...@vinciconsulting.com<mailto:pvi...@vinciconsulting.com>] -=-=-=-=-=-=-=-=-=-=-=-
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#12823): https://lists.fd.io/g/vpp-dev/message/12823 Mute This Topic: https://lists.fd.io/mt/31234917/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
