Bug 222347

http://www.redhat.com/archives/fedora-maintainers/2007-January/msg00017.html
http://koti.welho.com/vskytta/install-infos/g-wrap.patch

Thanks, fixed. -9 uploaded.

Note: the currently-in-development version of gnucash moves from g-wrap to swig
for its guile bindings; when that version is added to the development tree,
g-wrap will be orphaned, as nothing else uses it. I'll still maintain it for
older trees that need it.

rpmlint has some complaints:

This just looks like a typo in the changelog:
   W: g-wrap incoherent-version-in-changelog 1.9.6-10 1.9.6-9

I'm not really sure about these:
   W: g-wrap unused-direct-shlib-dependency 
    /usr/lib64/libgw-guile-standard.so.0.0.0 /lib64/libpthread.so.0
   W: g-wrap unused-direct-shlib-dependency 
    /usr/lib64/libgw-guile-gw-glib.so.0.0.0 /lib64/libglib-2.0.so.0
   W: g-wrap unused-direct-shlib-dependency 
    /usr/lib64/libgw-guile-gw-glib.so.0.0.0 /lib64/libpthread.so.0
I don't think they're especially problematic and I guess it's possible that
they're necessary.

I admit I find it odd to see /usr/lib64/g-wrap/include/ffi.h:
   E: g-wrap-devel only-non-binary-in-usr-lib
Why not /usr/share/g-wrap, or with the rest of the installed headers in
/usr/include/g-wrap?

Seems to me that the license is LGPL, not GPL.  The libffi license seems to be
MIT but that's compatible and doesn't change the overall license of the package.

I note that there is a test suite, but some notes about it being disabled on
x86_64.  I tried a quick "make check" on i386 rawhide and all the tests failed
without attempting to actually test anything, so I must be missing something.

Review:
* source files match upstream:
   ddb0e31d40581402d6d7045cce7cdc79e0bc0627831a4b12012f45703446d311  
   g-wrap-1.9.6.tar.gz
* package meets naming and versioning guidelines.
* specfile is properly named, is cleanly written and uses macros consistently.
* summary is OK.
* description is OK.
* build root is OK.
X license field matches the actual license.
* license is open source-compatible.
* license text included in package.
O latest version is 1.9.8, but there are issues preventing its use.
* BuildRequires are proper.
* compiler flags are appropriate.
* %clean is present.
* package builds in mock (development, x86_64).
* package installs properly
* debuginfo package looks complete.
* rpmlint is silent.
* final provides and requires are sane:
  g-wrap-1.9.6-9.x86_64.rpm
   libgw-guile-gw-glib.so.0()(64bit)
   libgw-guile-standard.so.0()(64bit)
   libgwrap-core-runtime.so.0()(64bit)
   libgwrap-guile-runtime.so.0()(64bit)
   g-wrap = 1.9.6-9
  =
   /sbin/ldconfig
   guile
   libglib-2.0.so.0()(64bit)
   libguile.so.17()(64bit)
   libgw-guile-gw-glib.so.0()(64bit)
   libgw-guile-standard.so.0()(64bit)
   libgwrap-core-runtime.so.0()(64bit)
   libgwrap-guile-runtime.so.0()(64bit)
   libpthread.so.0()(64bit)
   libpthread.so.0(GLIBC_2.2.5)(64bit)

  g-wrap-devel-1.9.6-9.x86_64.rpm
   g-wrap-devel = 1.9.6-9
  =
   /bin/sh
   /sbin/install-info
   g-wrap = 1.9.6
   guile-devel
   libgw-guile-gw-glib.so.0()(64bit)
   libgw-guile-standard.so.0()(64bit)
   libgwrap-core-runtime.so.0()(64bit)
   libgwrap-guile-runtime.so.0()(64bit)
   pkgconfig

* %check is not present.  Test suite doesn't seem to do much.
* shared libraries present; ldconfig called as appropriate.
* unversioned .so files are in the -devel package.
* owns the directories it creates.
* doesn't own any directories it shouldn't.
* no duplicates in %files.
* file permissions are appropriate.
* scriptlets are OK (ldconfig, install-info).
* code, not content.
* documentation is small, so no -docs subpackage is necessary.
* %docs are not necessary for the proper functioning of the package.
* headers are in the -devel subpackage.
* a pkgconfig file is present in -devel; pkgconfig dependency is there.
* no static libraries.
* no libtool .la files.

(In reply to comment #3)
> rpmlint has some complaints:
> I admit I find it odd to see /usr/lib64/g-wrap/include/ffi.h:
>    E: g-wrap-devel only-non-binary-in-usr-lib
> Why not /usr/share/g-wrap, or with the rest of the installed headers in
> /usr/include/g-wrap?

It's wordsize-specific - having it in /usr/share or /usr/include would cause
multilib conflicts.

> Seems to me that the license is LGPL, not GPL.  The libffi license seems to be
> MIT but that's compatible and doesn't change the overall license of the package.

Oops, fixed in CVS.

> I note that there is a test suite, but some notes about it being disabled on
> x86_64.  I tried a quick "make check" on i386 rawhide and all the tests failed
> without attempting to actually test anything, so I must be missing something.

The test suite has never seemed to work. I should have probably put more effort
into figuring out why.

So...

> * no libtool .la files.

These are re-added in the current package; see bug 238263. gnucash uses a
version of libtool_ltdl as a module loader, and it does not work without either
a) .la files b) .so files in the main package. As moving the .so files breaks
multilib development, it was simplest to put the .la files back.

Oh, I was reviewing the -9 package since that one was the last one mentioned in
this ticket and the one the spec link above points to.  Which version is
current?  Should I just get it from the old core CVS?

Just grab the latest from /cvs/pkgs.

OK, all that's really missing is some commentary in the spec as to why the .la
files are included and why those headers are under /usr/lib.  Heck, since this
has already been imported, I'll even commit some if you like.

But I don't see anything holding up this package, so

APPROVED

Comments added in CVS.

Package Change Request
======================
Package Name: g-wrap
New Branches: EL-4 EL-5

CVS done.

Can this ticket be closed now?

Sure.