Fedora-Review 0.2.0

[Pierre-Yves Chibon]

On Thu, 2012-08-02 at 08:23 -0600, Nathanael D. Noblet wrote:
> On 08/02/2012 12:02 AM, Mathieu Bridon wrote:
> > On Wed, 2012-08-01 at 23:52 -0600, Nathanael D. Noblet wrote:
> >> Where would you like bug reports?
> >>
> >> I tried it against one of my own review tickets. It found a number of
> >> issues however almost all of them except one was wrong.
> >>
> >> For example it complained of no clean section with a rm -rf %{buildroot}
> >> which the specfile contained, same message except in the install section
> >> etc.
> >
> > Are you sure it wasn't complaining that the specfile actually contained
> > those lines?
> >
> > The Fedora guidelines say those lines are not needed, and shouldn't be
> > there for new packages, unless the package maintainer wants to ensure
> > compatibility with EPEL 5.
> 
> Hmm I guess I mis-read it then.
> 
> [!]: MUST Buildroot is not present
> [!]: MUST Package does not run rm -rf %{buildroot} (or $RPM_BUILD_ROOT) 
> at the beginning of %install
> 
> When I read that, the ! tells me I failed that. So Package does not run 
> rm -rf %{buildroot} at the beginning of %install. I think the double 
> negative there is what threw me. I think putting the MUST part into the 
> actual description would help. For example
> 
> [!] Package MUST NOT run rm -rf %{buildroot} ...
> 
> Clearly tells me that I failed that because I am running the rm command. 
> Whereas a MUST preceding the line and the failure can be interpreted as 
> I failed because the package does not run rm -rf ... Not sure if you see 
> what I mean. Now that it is pointed out it does make more sense.
> 
> Thoughts? Is it just me that read that totally wrong?

Maybe it would be better formulated as:
[!]: MUST: Buildroot MUST NOT be present (NOTE: this is not true for
EPEL5)
[!]: MUST: Package MUST NOT run rm -rf %{buildroot} (or $RPM_BUILD_ROOT)
at the beginning of %install

What do you think?

Pierre