On Fri, Apr 03, 2020 at 10:18:02AM +0200, David Marchand wrote:
> On Fri, Apr 3, 2020 at 10:06 AM Pavan Nikhilesh Bhagavatula
> <pbhagavat...@marvell.com> wrote:
> >
> > ><david.march...@redhat.com> wrote:
> > >> On Thu, Apr 2, 2020 at 8:33 PM Jerin Jacob <jerinjac...@gmail.com>
> > >wrote:
> > >> - You can still find traces in meson files.
> > >> $ git grep -il allow.experimental.api |grep meson.build$ |grep -vE
> > >> '^((app|drivers|lib)/meson.build|examples/)'
> > >> app/test/meson.build
> > >> drivers/net/e1000/base/meson.build
> > >> drivers/net/enic/meson.build
> > >> drivers/net/i40e/base/meson.build
> > >> drivers/net/i40e/meson.build
> > >> drivers/net/ixgbe/base/meson.build
> > >> drivers/net/octeontx/base/meson.build
> > >> drivers/net/qede/base/meson.build
> > >> drivers/net/sfc/base/meson.build
> > >> drivers/net/thunderx/base/meson.build
> > >> lib/librte_mempool/meson.build
> > >
> > >Sorry, those drivers passing directly the cflag were special cases.
> >
> > Yup. Mostly base static libs.
>
> Actually, how about something like this in drivers/meson.build ?
> Bruce, do you see something wrong with this ?
>
> on top of Pavan patch:
>
> diff --git a/drivers/meson.build b/drivers/meson.build
> index 987e616ae7..4d8f842ab8 100644
> --- a/drivers/meson.build
> +++ b/drivers/meson.build
> @@ -20,7 +20,7 @@ dpdk_driver_classes = ['common',
> disabled_drivers = run_command(list_dir_globs, get_option('disable_drivers'),
> ).stdout().split()
>
> -default_cflags = machine_args
> +default_cflags = machine_args + ['-DALLOW_EXPERIMENTAL_API']
> if cc.has_argument('-Wno-format-truncation')
> default_cflags += '-Wno-format-truncation'
> endif
I see nothing wrong with that.
Out of interest, is there a reason we don't just set it at the top level as
a project c_arg, e.g. after setting "-include rte_config.h" in
config/meson.build?
