[Bug 630208] Review Request: ghc-csv - CSV loader and dumper

[bugzilla at redhat.com]

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


https://bugzilla.redhat.com/show_bug.cgi?id=630208

Ben Boeckel <mathstuf at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
               Flag|needinfo?(mathstuf at gmail.co |
                   |m)                          |

--- Comment #2 from Ben Boeckel <mathstuf at gmail.com> 2010-09-11 15:35:23 EDT ---
(In reply to comment #1)
> I am utterly unfamiliar with Haskell.  I'm happy to negotiate on any point,
> given my ignorance.
> 
> The strange permissions are exactly that.  Wouldn't 0644 make more sense?
> Since ghc-csv-prof is a profiling package, I assume that its errors and
> warnings are irrelevant.  A comment to this effect in the .spec would be nice.

The strange permissions are a relic of my umask of 027. When put into git and
the lookaside, they get the proper 644 (git only tracks +x or -x and the
lookaside has 644).

> [-] MUST: The spec file for the package MUST be legible.
> 
> There is apparently some pretty good voodoo in ghc-rpm-macros.  I'd like to see
> some comments describing that this is the reason for common_summary being
> defined buy only appearing once and the outright lack of %files and %doc - at
> least, I'm assuming that that's the reason.  As-is, I'd say that this .spec
> file is not legible to someone without an existing ghc background.

The spec magic is standard ghc-rpm-macros voodoo. The bulk of it is generated
by the cabal2spec tool.

The ghc_lib_package macro is in the following file:

   
http://pkgs.fedoraproject.org/gitweb/?p=ghc-rpm-macros.git;a=blob;f=ghc-rpm-macros.ghc

%files is generated by grepping the cabal file for the license file that gets
installed and adding it to %doc. It also makes the information for the -devel
and -prof (which is akin to a -debuginfo of sorts) packages. Haskell is
predictable since cabal standardizes many things and the directories and such
are generated given %{pkg_name} and %{version}.

> [?] MUST: Every binary RPM package (or subpackage) which stores shared library
> files (not just symlinks) in any of the dynamic linker's default paths, must
> call ldconfig in %post and %postun.
> 
> Is it fair to say that this doesn't apply to
> /usr/lib64/ghc-6.12.3/csv-0.1.1/libHScsv-0.1.1-ghc6.12.3.so?

The ghc_lib_package runs the ghc-pkg recache which registers the library with
ghc's linking.

> [?] MUST: A Fedora package must not list a file more than once in the spec
> file's %files listings.
> 
> I'm assuming that ghc-rpm-macros enforces this, but I have no easy way of
> knowing.  Legibility is not there.

rpm -qlp ghc-csv-*.<arch>.rpm

The COPYING file should be in there.

> [-] MUST: Permissions on files must be set properly. Executables should be set
> with executable permissions, for example. Every %files section must include a
> %defattr(...) line.
> 
> The 0640 permissions found by rpmlint are probably not "proper".  There is no
> legible %files section or comment.

> [?] MUST: Static libraries must be in a -static package.
> 
> ghc-csv-devel contains /usr/lib64/ghc-6.12.3/csv-0.1.1/libHScsv-0.1.1.a; I'm
> unsure if this is qualifies as a static library.

Haskell is a statically built language for some things. Hopefully ghc-7.0 will
help get us proper shared libraries across the board.

-- 
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.