The specfile references below are to the
c1de1c8d026610913b3a699545462bc4605d4ed4
version of
https://src.fedoraproject.org/rpms/cppad/blob/rawhide/f/cppad.spec
This runs fine (for me) with the `fedpkg mockbuild` command
On 2/24/21 6:53 AM, Dominik 'Rathann' Mierzejewski wrote:
... snip ...
I also took a look at the spec file. Can you explain why you're patching
the source with sed in %prep section instead of using patches (unified
diff files) or, even simpler, grabbing the latest upstream release
tarball? I can see 20210000.5 here:
https://github.com/coin-or/CppAD/releases
I have uploaded the 20210000.5 source and simplified the specfile above.
You should also make the Provides: here
https://src.fedoraproject.org/rpms/cppad/blob/rawhide/f/cppad.spec#_40
archful (i.e. add %{?_isa}) as the subpackage.
You could simplify things by dropping the -devel subpackage and adding
Provides: %{name}-devel%{?_isa} = %{version}-%{release}
to the main package instead. Remember to add appropriate Obsoletes:
then, too.
Would you please point me to an example spec file that uses %(?_isa) in its Provides
section.
In addition, an example that uses Provides for %{name}-devel would be helpful.
In %install, rm -rf %{buildroot} is completely unnecessary since
RHEL/EPEL5 and make install DESTDIR=%{buildroot} can be replaced with
%make_install . Similarly, make %{?_smp_flags} can be replaced with
%make_build .
This has been done in the specfile above.
Finally, why aren't you using the %setup macro in %prep?
When I try to add the line `%setup -q` directly after the line `%prep`
(in the specfile above) I get errors. Here is what happens when I try `fedpkg lint`:
cppad>fedpkg lint
error: line 162: %package debuginfo
: package cppad-debuginfo already exists
error: query of specfile /home/bradbell/fedora/cppad/cppad.spec failed, can't
parse
Could not execute lint: Could not get n-v-r-e from
/home/bradbell/fedora/cppad/cppad.spec
The are other changes that need to be made to remove my manually doing what the `%setup
-q` should
do, but I cannot get passed the error above.
Your SourceN lines should start with 0 instead of 1.
This has been done in the specfile above.