[Bug 450323] Review Request: coq - Coq proof management system

bugzilla at redhat.com bugzilla at redhat.com
Mon Jun 9 09:09:43 UTC 2008


Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug report.

Summary: Review Request: coq - Coq proof management system


https://bugzilla.redhat.com/show_bug.cgi?id=450323





------- Additional Comments From rjones at redhat.com  2008-06-09 05:09 EST -------
These bits of the spec file are all wrong:

  # Test for emacs site_lisp directory, if so, add relevant files to roster, else, don't try and install
  ...
  # Test for tex directory, if so, add relevant files to roster, else, don't try and install
  ...

It's not acceptable to have different RPMs being produced depending on what
happens to be installed at the time.  Instead, assume those directories / packages
are installed and ensure this by having a complete list of BuildRequires.

Your BuildRequires is missing at least emacs, texlive-latex, another texlive-*
package which provides /usr/share/texmf/tex/latex/misc (I couldn't find
which one).

Once you think you've got a complete list of BuildRequires, you should then
scratch-build the package in koji:

  koji build --scratch dist-f10 coq-8.1pl3-1.fc9.src.rpm

This will almost certainly fail, but it should fail in a way which tells you which
extra BuildRequires you are missing and any other problems that you'll
encounter in the real build.

When you have a successful scratch-build in Koji, please attach a link to
the Koji build here.

Next thing you should do is to run rpmlint on all the RPMs (source and binary
RPMs).  rpmlint output should be nil for this package.

Another thing I notice in the spec file:

  %{_bindir}/parser
  %{_bindir}/parser.opt
  # I suppose technically we might not have built parser.opt, but my efforts to fix this problem re: accounting for this in 
the file manifest have failed

This is against the OCaml packaging policy which requires that you package the
best possible binary (ie. native, if available, else bytecode).  You can easily do
this by testing for the presense of ocamlopt.  See the first line of our sample
specfile:

  http://fedoraproject.org/w/uploads/5/5c/Packaging_OCaml_ocaml-foolib.spec

-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.




More information about the Fedora-package-review mailing list