[Libosinfo] [osinfo-db-tools PATCH 2/8] Add support to meson build system

Pavel Hrdina phrdina at redhat.com
Thu Jun 20 18:00:16 UTC 2019


On Thu, Jun 20, 2019 at 06:24:34PM +0100, Daniel P. Berrangé wrote:
> On Thu, Jun 20, 2019 at 07:10:54PM +0200, Fabiano Fidêncio wrote:
> > On Thu, Jun 20, 2019 at 7:05 PM Daniel P. Berrangé <berrange at redhat.com> wrote:
> > >
> > > On Thu, Jun 20, 2019 at 06:42:29PM +0200, Pavel Hrdina wrote:
> > > > Few more notes that I've noticed:
> > > >
> > > >     - There is a regression, with meson we no longer have generated RPM
> > > >       spec files in tarball.
> > >
> > > Yep, that's desirable so users can do  "rpmbuild -ta TARBALL", though
> > > note only the native spec is desired - not the generated mingw spec.
> > >
> > > >     - With autotools there were some configure options:
> > > >         --enable-vala [default check]
> > > >         --enable-coverage [default no]
> > > >         --with-usb-ids-path [default (internal)]
> > > >         --with-pci-ids-path [default (internal)]
> > > >         --enable-werror [default git ? yes : no ]
> > >
> > > Yep, we should keep at least the last 3 args. The coverage stuff coukd
> > > simply be auto-detected with no arg.  I'm on the fence wrt whether we
> > > need a vala arg or not
> > 
> > No, we shouldn't.
> > ```
> > fidencio at laerte /tmp/osinfo-db-tools $ ./configure --enable-vala
> > --enable-coverage --with-usb-ids-path --with-pci-ids-path
> > --enable-werror
> > configure: WARNING: unrecognized options: --enable-vala,
> > --enable-coverage, --with-usb-ids-path, --with-pci-ids-path
> > ...
> > configure: WARNING: unrecognized options: --enable-vala,
> > --enable-coverage, --with-usb-ids-path, --with-pci-ids-path
> > ```
> 
> Oh right, this was leftover cruft that was in the libosinfo code
> originally that we mistakenly didn't delete.

I see, should have checked whether any of the options in configure.ac
are actually used.

> > About the --enable-werror, it's just a matter of passing -werror to meson.
> 
> but this requires a developer opt-in - IMHO we really want to keep having
> werror always enabled when building from .git at least. We can do that
> without a flag if we're ok not providing a way to disable it when a .git
> dir exists.

Agreed, we should have werror enabled by default for git builds, I don't
have any preference whether we should make it configurable or not.

Pavel
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/libosinfo/attachments/20190620/58913714/attachment.sig>


More information about the Libosinfo mailing list