[PATCH] Revert "spec: Simplify setting features off by default"

Andrea Bolognani abologna at redhat.com
Wed Oct 28 21:18:02 UTC 2020

On Wed, 2020-10-28 at 16:55 -0400, Neal Gompa wrote:
> On Tue, Oct 27, 2020 at 9:30 AM Andrea Bolognani <abologna at redhat.com> wrote:
> > On Tue, 2020-10-27 at 14:05 +0100, Andrea Bolognani wrote:
> > > Note that in this case I've removed
> > > 
> > >   # fuse is used to provide virtualized /proc for LXC
> > >   %if %{with_lxc}
> > >       %define with_fuse      0%{!?_without_fuse:1}
> > >   %endif
> > > 
> > > from the spec to make sure that the value for the 'fuse' option
> > > passed to Meson depended solely on the value of the _without_fuse
> > > macro, and then checked the rpmbuild output to compare.
> Ugh, you're right, and those values need to be changed to 1.

Yeah, maybe we should reconsider whether the features in our spec
which are off by default should not be on by default instead. It's
not something that I've tried to do with my previous patches in the
area: I've limited myself to fixing logic issues and cleaning up the
implementation, leaving the semantics unmodified.

> > Also note that I'm aware you want to eventually push for adoption of
> > the standard bcond macros, and I fully stand behind that desire! If
> > this patch had been the first in a series that introduced bcond
> > support and was clearing the path for that, I would have zero
> > problems with it. As it is, however, you're simply reintroducing some
> > of the obfuscation we had recently managed to get rid of, without
> > getting anything in return.
> Fixing this so that I can switch to bconds is going to be a massive
> rewrite of how feature enablement works. That is not something I can
> push for a 6.9.0 freeze break patch.
> My in-progress rewrite is going to be a massive break in how this
> is managed...

The kind of work you're describing is definitely not freeze material!
It's much better to merge big series early in the development cycle

In any case, since you agree with me that your revert left the logic
unchanged and merely reintroduced the old obfuscation, I'm going to
go ahead and... Revert it :)

Looking forward to your bcond patches!

Andrea Bolognani / Red Hat / Virtualization

More information about the libvir-list mailing list