BTW this is a reason why build fails in covscan

@rcritten,
Do you have a time to review this PR?

Yes I'm going to look at it when the tests are fixed. It seemed pointless to push this given that the build will fail later due to make check issues.

This test is actually wrong. On a non-RHEL system this will be 0 which is < 6 so will succeed.

I'm not sure how far we want to take it but dropping support for Fedora 15, RHEL < 6, etc may be the best way to resolve this.

Otherwise this could be changed to something like

%if 0%{?rhel} < 6 && 0%{?rhel} >= 5

BTW, I don't mean to imply the wrong test is due to you, it's always been there and nobody noticed :-(

BTW, I don't mean to imply the wrong test is due to you, it's always been there and nobody noticed :-(

I used a little bit different version %if 0%{?rhel} && 0%{?rhel} <= 6
It is not ideal because it does not cover fedora :-) And these version of fedora are really ancient.

initially, I just wanted remove that BR but I noticed that conditional and moved it there. I did/do not want to drop unsupported distros from spec file. I prefer minimal change.

We are in agreement with the minimal changes. Doing otherwise would require touching the systemd and sysvinit variables and the change gets quite a bit more complex (though the spec quite a bit less).

Right, as a side-effect I'll need to ensure this change doesn't affect shipping versions of Fedora since it would have had that BR as a side-effect but in principle your change looks good to me.

Can you update the patch with your proposed conditional? I'll merge it with that.

Sorry I forgot to force push changes :-)