Burt and Ole, On Sun, Mar 19, 2017 at 7:26 AM, <otr...@employees.org> wrote: > >> >> Hmm, so https://gerrit.fd.io/r/#/c/5781/ >> isn't sufficient? >> > Necessary, yes. Sufficient? No.
On the other hand... On Sun, Mar 19, 2017 at 11:51 AM, Burt Silverman <bur...@gmail.com> wrote: > Ugh, I made a terrible bone head mistake... Possibly I never ran make > install-dep. Even if I had, I was unaware that it is a good idea to run it > again, just to be sure, in a case like this. I probably thought it was like > make bootstrap, where running a 2nd time doesn't help. Apologies to Ed and > Ole for misinformation. So, Jon, were you in the same boat with me -- > didn't do a double check of make install-dep? I guess so, because you still > had the problem after Ole's fix. > > Burt > This was precisely the problem here. I'll try to say this as politely as I can... Wow. That's some blind-siding sh*t. So, let's talk about that a bit. First, thank you for identifying the issue! This does indeed fix the build locally, and bring us back to online par. Thank you! Second, the notion of requiring repeated running of the make install-dep target as part of our daily build process from our CI engine is just not going to happen. NFW. We're not running anything has root like that. It's a bad idea for many reasons. On the flip side, I can make a job that "notices" a change in the installed packaged requirement and run that as, say, a daily job and incidentally notice that updates are needed. Sure, polling like that sucks; an interrupt here with a simple "Heads up! The install-deps have changed" would have been awesome! Thanks, jdl
_______________________________________________ vpp-dev mailing list vpp-dev@lists.fd.io https://lists.fd.io/mailman/listinfo/vpp-dev
