Hi Luis, On Fri, May 13, 2022 at 12:38:30PM -0700, Luis Chamberlain wrote: > Two playbooks share the concept of git cloning kdevops into > the target nodes (guests, cloud hosts, baremetal hosts) so that > expunge files can be used for avoiding tests. If you decide > you want to change the URL for that git tree it may not be > so obvious what to do. > > Fortunately the solution is simple. You just tell ansible to use > the new git tree URL. That's it. It won't remove the old directory > and things work as expected. > > But since we use the kdevops git tree on both fstests and blktests > it is not so obvious to developers that the thing to do here is > to just run 'make fstests' or 'make blktests' and even that is not > as efficient as that will also re-clone the fstests or blktests > tree respectively. When we just want to reset the kdevops git tree > we currently have no semantics to specify that. But since this is > a common post-deployment goal, just add a common playbook that let's > us do common tasks. > > All we need then is the kconfig logic to define when some commmon > tasks might make sense. So to reset your kdevops git tree, all you > have to do now is change the configuration for it, then run: > > make > make kdevops-git-reset > While I do like the idea of having this option, I still do not understand the main use case to have it as a separate make target. Wouldn't the developer already put the custom kdevops tree with CONFIG_WORKFLOW_KDEVOPS_GIT during the initial make menuconfig phase? I am just trying to understand the usecase when someone wants to change the kdevops tree after a test run. Maybe I am missing something here. -- Pankaj Raghav
