Bug 501958 - Review Request: scotch - Graph, mesh and hypergraph partitioning library
Summary: Review Request: scotch - Graph, mesh and hypergraph partitioning library
Keywords:
Status: CLOSED NEXTRELEASE
Alias: None
Product: Fedora
Classification: Fedora
Component: Package Review
Version: rawhide
Hardware: All
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Jason Tibbitts
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2009-05-21 13:14 UTC by Deji Akingunola
Modified: 2014-12-01 13:11 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2009-11-07 15:10:07 UTC
Type: ---
Embargoed:
j: fedora-review+
gwync: fedora-cvs+


Attachments (Terms of Use)

Description Deji Akingunola 2009-05-21 13:14:06 UTC
Spec URL: http://deji.fedorapeople.org/scotch.spec
SRPM URL: http://deji.fedorapeople.org/scotch-5.1.6-1.fc11.src.rpm
Description: Scotch is a software library for graph and mesh/hypergraph partitioning and sparse matrix ordering.

Comment 1 Jason Tibbitts 2009-06-04 02:43:45 UTC
This does not build.  A scratch build: http://koji.fedoraproject.org/koji/taskinfo?taskID=1392846

Please do a local mock build or a koji scratch build when submitting a package.

Comment 2 Deji Akingunola 2009-06-04 09:07:11 UTC
(In reply to comment #1)
> This does not build.  A scratch build:

Fixed.
Spec URL: http://deji.fedorapeople.org/scotch.spec
SRPM URL: http://deji.fedorapeople.org/scotch-5.1.6-2.fc11.src.rpm

Comment 3 Jason Tibbitts 2009-06-11 00:01:33 UTC
A few additional comments.  Generally I would suggest that you run rpmlint on your packages and fix the obvious things.

CeCILL-C_V1-*.txt need to be run through iconv.

The file in ld.so.conf.d either shouldn't be marked %config or should be marked %config(noreplace).  I don't think the user is supposed to edit it, so the former seems more correct to me.

For some weird reason all of the libraries and executables end up mode 0775 for me.  Do you see that?  It could be a weird problem with my buildsystem, but I'm not sure.

No parallel make?  If it doesn't work, you should at least add a comment to that effect.

Several of the executable names seem destined to conflict.  For example, /usr/bin/gbase conflicts with http://www.hibernaculum.net/gbase/index.php which isn't in Fedora but which is in various other distros.  /usr/bin/mtst conflict with the mt-st package in some distros (suse, for example), although Fedora seems to call the binary simply /usr/bin/mt.  /usr/bin/gpart conflicts with the gpart package currently in Fedora (i586 only, though).

Comment 4 Deji Akingunola 2009-06-12 10:13:13 UTC
I have fixed issues pointed out above; run the license files through iconv, drop the %config mark on the file in ld.so.conf.d, and use parallel make.

I've also contacted the upstream author about the executable names, and he is OK with me prefixing them all by scotch_ (I believe he is going to apply the same change upstream with the next release of the library).
Thanks.

Spec URL: http://deji.fedorapeople.org/scotch.spec
SRPM URL: http://deji.fedorapeople.org/scotch-5.1.6-3.fc11.src.rpm

Comment 5 Jason Tibbitts 2009-07-05 23:31:55 UTC
This builds fine, though I'm still seeing the 0775 permissions thing.  You didn't respond to tell me if you see that as well, so I did a koji build and it doesn't happen on the buildsys, so I'll just assume some problem with mock on my end.  Ignoring those permission complaints, I get:
  scotch-devel.x86_64: W: no-documentation
  scotch-static.x86_64: W: no-documentation               
These aren't problems.

  scotch.x86_64: W: non-conffile-in-etc /etc/ld.so.conf.d/scotch.conf
  scotch-devel.x86_64: W: only-non-binary-in-usr-lib
These are necessary because of the way the libraries from this package are placed into a subdirectory.  Honestly, looking at the contents of /usr/lib64/scotch, I have to wonder why you'd want to go through that extra effort.  There's nothing there that's going to conflict, and you're not doing any fancy versioning of the directory that you'd need for parallel installation or anything like that, so why not just put the libraries in %_libdir and dispense with the ld.so.conf.d magic?

The source file downloaded from the Source0 URL is not the same as the source file in the tarball.  The download is quite a bit smaller, and seems to contain older files.

CeCILL-C is not GPL-compatible, but mpich2 is MIT and zlib is very liberally licensed so I see no linking issues.

I note that the -static subpackage has no dependency on the -devel package.  I don't believe this is mandatory, but most -static packages seem to do it and it does make sense from the perspective of someone who wants to use it (yum install scotch-static would bring in what you need to actually use those libraries).

X source files do not match upstream.
* package meets naming and versioning guidelines.
* specfile is properly named, is cleanly written and uses macros consistently.
* summary is OK.                                                              
* description is OK.                                                          
* dist tag is present.
* build root is OK.
* license field matches the actual license.
* license is open source-compatible.
* license text included in package.
* latest version is being packaged.
* BuildRequires are proper.
* compiler flags are appropriate.
* %clean is present.
* package builds in mock (rawhide, x86_64).
* package installs properly.
* debuginfo package looks complete.
? rpmlint has a couple of complaints which may go away.
* final provides and requires:
  scotch-5.1.6-3.fc12.x86_64.rpm
   libptscotch.so.0()(64bit)       
   libptscotcherr.so.0()(64bit)    
   libptscotcherrexit.so.0()(64bit)  
   libptscotchparmetis.so.0()(64bit)  
   libscotch.so.0()(64bit)            
   libscotcherr.so.0()(64bit)         
   libscotcherrexit.so.0()(64bit)     
   libscotchmetis.so.0()(64bit)       
   scotch = 5.1.6-3.fc12              
   scotch(x86-64) = 5.1.6-3.fc12      
  =                                
   /sbin/ldconfig                     
   libgfortran.so.3()(64bit)          
   libmpich.so.1.1()(64bit)           
   libptscotch.so.0()(64bit)          
   libptscotcherr.so.0()(64bit)       
   libptscotcherrexit.so.0()(64bit)   
   libptscotchparmetis.so.0()(64bit)  
   libscotch.so.0()(64bit)
   libscotcherr.so.0()(64bit)
   libscotcherrexit.so.0()(64bit)
   libscotchmetis.so.0()(64bit)
   libz.so.1()(64bit)

  scotch-devel-5.1.6-3.fc12.x86_64.rpm
   scotch-devel = 5.1.6-3.fc12
   scotch-devel(x86-64) = 5.1.6-3.fc12
  =
   libptscotch.so.0()(64bit)
   libptscotcherr.so.0()(64bit)
   libptscotcherrexit.so.0()(64bit)
   libptscotchparmetis.so.0()(64bit)
   libscotch.so.0()(64bit)
   libscotcherr.so.0()(64bit)
   libscotcherrexit.so.0()(64bit)
   libscotchmetis.so.0()(64bit)
   scotch = 5.1.6-3.fc12

  scotch-static-5.1.6-3.fc12.x86_64.rpm
   scotch-static = 5.1.6-3.fc12
   scotch-static(x86-64) = 5.1.6-3.fc12
  =
   scotch = 5.1.6-3.fc12

* %check is present; no test suite upstream.  I've no way to test this.  The 
  executables don't crash, but I have no date to feed them.
* shared libraries are installed:
  ldconfig called properly
  unversioned .so files are in the -devel subpackage.
* owns the directories it creates.
* doesn't own any directories it shouldn't.
* no duplicates in %files.
* file permissions are appropriate.
* no generically named files
* scriptlets OK (ldconfig).
* code, not content.
* documentation is small, so no -doc subpackage is necessary.
* %docs are not necessary for the proper functioning of the package.
* headers are in the -devel subpackage.
* no pkgconfig files.
* static libraries present in a separate -static package.
* no libtool .la files.

The package review process needs reviewers!  If you haven't done any package
reviews recently, please consider doing one.

Comment 6 Jason Tibbitts 2009-11-03 15:43:15 UTC
Four months with no response to the above review commentary; I'll close this out soon if there's no further progress.

Comment 7 Deji Akingunola 2009-11-03 16:45:28 UTC
I'm so sorry for the lack of update on the bug, it has somehow fallen off my radar.

I've implemented your suggestions of putting the libs directly under $libdir, and made the -static subpackage require the -devel subpackage. I've also updated to the latest upstream version (v_5.1.7) thus solving the tarball mdsum mismatch issue (upstream must have subtly changed it).
Thanks.

Spec URL: http://deji.fedorapeople.org/scotch.spec
SRPM URL: http://deji.fedorapeople.org/scotch-5.1.7-1.fc12.src.rpm

Comment 8 Jason Tibbitts 2009-11-04 18:27:38 UTC
I still seem to be getting a different tarball than what's in the package, and the differences are quite significant (18000 lines of patch between what's in your src.rpm and the one I got with spectool -g scotch.spec.  To double check, I manually did the substitution in the Source0: line and fetched it via wget and links and what I downloaded was still different.  I think we need to get to the bottom of this before this package can be accepted.

It looks like all of the other issues are resolved.

Comment 9 Deji Akingunola 2009-11-04 18:50:01 UTC
My fault; I forgot to update the source url. Done than now, and hopefully fix the issue.

Spec URL: http://deji.fedorapeople.org/scotch.spec
SRPM URL: http://deji.fedorapeople.org/scotch-5.1.7-2.fc11.src.rpm

Comment 10 Jason Tibbitts 2009-11-04 18:55:50 UTC
Oh, hmm, their system does make it a bit difficult to figure that out, doesn't it?

Indeed, everything looks good now.

APPROVED

Comment 11 Deji Akingunola 2009-11-05 02:34:29 UTC
New Package CVS Request
=======================
Package Name: scotch
Short Description: Graph, mesh and hypergraph partitioning library
Owners: deji
Branches: F-10 F-11 F-12
InitialCC:

Comment 12 Kevin Fenzi 2009-11-06 20:37:13 UTC
cvs done.

Comment 13 Deji Akingunola 2009-11-07 15:10:07 UTC
Imported and built in all branches. Thanks Jason for the review.

Comment 14 Sandro Mani 2014-11-28 12:03:26 UTC
Package Change Request
======================
Package Name: scotch
New Branches: el6 epel7
Owners: smani
InitialCC:

Comment 15 Gwyn Ciesla 2014-12-01 13:11:05 UTC
Git done (by process-git-requests).


Note You need to log in before you can comment on or make changes to this bug.