On 08/10/20 11:17, misono.tomoh...@fujitsu.com wrote: > Hi Paolo > > Thanks a lot for the clear explanation. I will update the patch to follow the > meson style. > I realized virtiofsd actually needs tools (i.e. "--disable-tools > --enable-virtiofsd" > does not work with above meson.build) since virtiofsd requires libvhost_user > which will > be built ony when tools are built. So, I will keep the current dependency > check (except 'have_system').
I'm thinking the behavior for --disable-tools --enable-virtiofsd would be confusing. Therefore, another possibility is not introducing --enable-virtiofsd. Instead, you can reuse --enable-vhost-user-fs: - --enable-vhost-user-fs will fail if tools are enabled and cap-ng or seccomp are unavailable - --enable-vhost-user-fs --disable-tools will not look for cap-ng or seccomp, because then the flag only controls inclusion of vhost-user-fs in the emulators - if "--enable-vhost-user-fs" is not specified and tools are enabled, vhost-user-fs will only be included in the emulators if cap-ng and seccomp are available - if "--enable-vhost-user-fs" is not specified, tools are enabled and cap-ng/seccomp are unavilable, vhost-user-fs will not be included in the emulators either - if "--enable-vhost-user-fs" is not specified but tools are not enabled, configure will not check if cap-ng or seccomp. In this case reusing most of your previous patch, and not moving everything to meson, is totally okay. I don't want you to impose more transition work. Paolo
