[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #36 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.el6 has been pushed to the Fedora EPEL 6 stable repository. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=pILNk3wRrWa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Fedora Update System upda...@fedoraproject.org changed: What|Removed |Added Status|ON_QA |CLOSED Resolution|--- |ERRATA Last Closed||2013-04-04 19:51:13 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=olLBtsqgp6a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #34 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.fc18 has been pushed to the Fedora 18 stable repository. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=zUxsSIiBvaa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #35 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.fc17 has been pushed to the Fedora 17 stable repository. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=d8OOneIQLQa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Fedora Update System upda...@fedoraproject.org changed: What|Removed |Added Status|MODIFIED|ON_QA -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=gQDVoyuRnYa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #33 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.el6 has been pushed to the Fedora EPEL 6 testing repository. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=7pM2jDL46ca=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Fedora Update System upda...@fedoraproject.org changed: What|Removed |Added Status|NEW |MODIFIED -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=6YE0jZos78a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #30 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.fc17 has been submitted as an update for Fedora 17. https://admin.fedoraproject.org/updates/MUMPS-4.10.0-9.fc17 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=TsAjWL9Ka7a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #31 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.fc18 has been submitted as an update for Fedora 18. https://admin.fedoraproject.org/updates/MUMPS-4.10.0-9.fc18 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=b60n6hLjaJa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #32 from Fedora Update System upda...@fedoraproject.org --- MUMPS-4.10.0-9.el6 has been submitted as an update for Fedora EPEL 6. https://admin.fedoraproject.org/updates/MUMPS-4.10.0-9.el6 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=0OhpoYoxOya=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Antonio Trande anto.tra...@gmail.com changed: What|Removed |Added Blocks|894604 | -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=AfGoCqIWkHa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #29 from Pavol Babinčák pbabi...@redhat.com --- Git done (by process-git-requests). -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=4qMDmp4wtza=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Antonio Trande anto.tra...@gmail.com changed: What|Removed |Added Flags||fedora-cvs? -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=r75KXG864Sa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #28 from Antonio Trande anto.tra...@gmail.com --- New Package SCM Request === Package Name: MUMPS Short Description: A MUltifrontal Massively Parallel sparse direct Solver Owners: sagitter Branches: f17 f18 f19 el6 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=JHBeP5TKbYa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com changed: What|Removed |Added Flags||fedora-review? -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=Utgc1ruqCda=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com changed: What|Removed |Added Blocks|177841 (FE-NEEDSPONSOR) | Flags|fedora-review? | Flags||fedora-review+ --- Comment #27 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com --- Antonio, welcome as a new Fedora packager :-) Please check https://fedoraproject.org/w/index.php?title=Join_the_package_collection_maintainers#Add_Package_to_Source_Code_Management_.28SCM.29_system_and_Set_Owner for the next steps, starting with requesting MUMPS to be added to git, create branches if you wish to make updates, e.g. build it also for f18. Feel free to ask me for any help in these initial steps. The MUMPS package is APPROVED. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=DWoDLxLRtYa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #21 from Antonio Trande trp...@katamail.com --- (In reply to comment #20) No need to %post and %postun in the devel subpackage; ldconfig should be run only in %post* of the actual library package. Quote: If the package has multiple subpackages with libraries, each subpackage should also have a %post/%postun section that calls /sbin/ldconfig. http://fedoraproject.org/wiki/Packaging:Guidelines#Shared_Libraries I suppose that, in this case, it is useless since every sub-package depends by main one, so if I remove one of them, all will be removed and %post* will run at one time. Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-8.fc18.src.rpm Will you use this package for your 'coin-or-Ipopt' package ? Is it still necessary keep open my review request 908089 ? :) -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=EtTbKpTiwXa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #22 from Michael Schwendt mschwe...@gmail.com --- If the package has multiple subpackages with libraries, each subpackage should also have a %post/%postun section that calls /sbin/ldconfig. Each (sub-)package that stores shared libs in run-time linker's (!) search path should have a %post/%postun section that calls ldconfig. You can play with ldconfig -v to see what it does after adding/removing shared libs. The -devel package contains no real libs but just softlinks. ldconfig doesn't care about those links. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=CwgwaLgyFma=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #23 from Michael Schwendt mschwe...@gmail.com --- since every sub-package depends by main one, so if I remove one of them, all will be removed and %post* will run at one time. It doesn't work like that. If you remove a subpackage that depends on the base package, the base package stays installed. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=FAiT4cDmZ6a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #24 from Antonio Trande trp...@katamail.com --- Thank you Michael for your clarification. :) -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=jSHOym0jVra=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com changed: What|Removed |Added Blocks||894604 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=XiZBPu5ae6a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #25 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com --- Antonio, I made my coin-or-Ipopt package work with your MUMPS package, it did require a bit of patching and several experimental builds to get it to work with an external, and parallel, not bundled, sequential MUMPS. See https://bugzilla.redhat.com/show_bug.cgi?id=894604 Well, I believe we should cooperate here, but not sure if it is not cheating if I give you the coin-or-Ipopt package so that I can review it :-) Either way, I think you can at least test your ascend package with your MUMPS package and my coin-or-Ipopt for now. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=lvfHKtXJuba=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #26 from Antonio Trande trp...@katamail.com --- (In reply to comment #25) Antonio, I made my coin-or-Ipopt package work with your MUMPS package, it did require a bit of patching and several experimental builds to get it to work with an external, and parallel, not bundled, sequential MUMPS. See https://bugzilla.redhat.com/show_bug.cgi?id=894604 Good! I close my Bug908089 and we can focus ourselves on coin-or-Ipopt. Well, I believe we should cooperate here, but not sure if it is not cheating if I give you the coin-or-Ipopt package so that I can review it :-) Waiting for someone else opinion. :) Either way, I think you can at least test your ascend package with your MUMPS package and my coin-or-Ipopt for now. Okay. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=ACRQK3k8SHa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Bug 913152 depends on bug 908089, which changed state. Bug 908089 Summary: Review Request: ipopt - Large-scale optimisation solver https://bugzilla.redhat.com/show_bug.cgi?id=908089 What|Removed |Added Status|NEW |CLOSED Resolution|--- |DUPLICATE -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=pIEFA2mjiba=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #20 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com --- Only 2 minor issues: Add an ending . to the description of the devel and examples packages. No need to %post and %postun in the devel subpackage; ldconfig should be run only in %post* of the actual library package. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=BbFTuc196sa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #18 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com
---
Almost there, should be close to done now :-)
1. Development (unversioned) .so files in -devel subpackage, if present.
Note: Unversioned so-files directly in %_libdir.
---
You got it reversed :-)
lib*-%{version}.so should go in the main package, and
lib*.so should go in the -devel package, example:
$ file /usr/lib64/libcmumps.so
/usr/lib64/libcmumps.so: symbolic link to `/usr/lib64/libcmumps-4.10.0.so'
$ rpm -qf /usr/lib64/libcmumps.so /usr/lib64/libcmumps-4.10.0.so
MUMPS-4.10.0-6.fc19.x86_64
MUMPS-devel-4.10.0-6.fc19.x86_64
2. Large documentation must go in a -doc subpackage.
Note: Documentation size is 1669120 bytes in 6 files.
---
I suggest not installing the .ps file, and then up to you if
creating or not a -doc package just for the .pdf. Since this
packge is expected to be a dependency of others, probably a
good idea do create a -doc package.
The ChangeLog file should go to the -devel package also.
3. SourceX / PatchY prefixed with %{name}.
Note: Patch0 (examples-mpilibs.patch) Patch1 (shared-pord.patch) Source1
(Makefile.par.inc) Patch2 (shared-mumps.patch)
---
You may rename the sources to prefix it with MUMPS-, but it is
optional (just to reduce the number the warnings)
Extra:
1. Add openssh-clients to BuildRequires, optionally making all the
%check conditional. Google'ing a bit I found the reason it would
fail by missing ssh or rsh:
http://www.open-mpi.org/community/lists/users/2010/07/13503.php
2. About the undefined-non-weak-symbol and unused-direct-shlib-dependency,
unless someone else has some comments, I believe it is ok, as the
package is functional. And it appears bogus, for example. it tells:
W: unused-direct-shlib-dependency /usr/lib64/libsmumps-4.10.0.so
libmpi_f77.so.1
and then
W: undefined-non-weak-symbol /usr/lib64/libdmumps-4.10.0.so mpi_reduce_
but the mpi_reduce_ symbol is defined in libmpi_f77.so.1
The warning should be either because of the mpi libraries not in
LD_LIBRARY_PATH at rpmlint runtime, or maybe (unlikely) because of the
reversed links in the main package and libraries in the -devel one
as explained above.
Package Review
==
Key:
[x] = Pass
[!] = Fail
[-] = Not applicable
[?] = Not evaluated
[ ] = Manual review needed
Issues:
===
[!]: Development (unversioned) .so files in -devel subpackage, if present.
Note: Unversioned so-files directly in %_libdir.
See: http://fedoraproject.org/wiki/Packaging/Guidelines#DevelPackages
[!]: Large documentation must go in a -doc subpackage.
Note: Documentation size is 1669120 bytes in 6 files.
See: http://fedoraproject.org/wiki/Packaging/Guidelines#PackageDocumentation
= MUST items =
C/C++:
[x]: Header files in -devel subpackage, if present.
[x]: Package does not contain any libtool archives (.la)
[x]: Package does not contain kernel modules.
[x]: Package contains no static executables.
[x]: Rpath absent or only used for internal libs.
[!]: Development (unversioned) .so files in -devel subpackage, if present.
Note: Unversioned so-files directly in %_libdir.
Generic:
[x]: Package is licensed with an open-source compatible license and meets
other legal requirements as defined in the legal section of Packaging
Guidelines.
[x]: Package successfully compiles and builds into binary rpms on at least one
supported primary architecture.
[x]: %build honors applicable compiler flags or justifies otherwise.
[x]: All build dependencies are listed in BuildRequires, except for any that
are listed in the exceptions section of Packaging Guidelines.
[x]: Package contains no bundled libraries.
[x]: Changelog in prescribed format.
[x]: Package does not run rm -rf %{buildroot} (or $RPM_BUILD_ROOT) at the
beginning of %install.
[x]: Sources contain only permissible code or content.
[x]: Each %files section contains %defattr if rpm 4.4
[x]: Macros in Summary, %description expandable at SRPM build time.
[x]: Package contains desktop file if it is a GUI application.
[x]: Development files must be in a -devel package
[x]: Package requires other packages for directories it uses.
[x]: Package uses nothing in %doc for runtime.
[x]: Package is not known to require ExcludeArch.
[x]: Package does not contain duplicates in %files.
[x]: Permissions on files are set properly.
[x]: Fully versioned dependency in subpackages, if present.
Note: No Requires: %{name}%{?_isa} = %{version}-%{release} in %package
devel, %package examples
[x]: Package complies to the Packaging Guidelines
[x]: Spec file lacks Packager, Vendor, PreReq tags.
[x]: If (and only if) the source package includes the text of the license(s)
in its own file, then that file, containing the text of the license(s)
for the package
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #19 from Antonio Trande trp...@katamail.com ---
(In reply to comment #18)
Almost there, should be close to done now :-)
1. Development (unversioned) .so files in -devel subpackage, if present.
Note: Unversioned so-files directly in %_libdir.
---
You got it reversed :-)
lib*-%{version}.so should go in the main package, and
lib*.so should go in the -devel package, example:
$ file /usr/lib64/libcmumps.so
/usr/lib64/libcmumps.so: symbolic link to `/usr/lib64/libcmumps-4.10.0.so'
$ rpm -qf /usr/lib64/libcmumps.so /usr/lib64/libcmumps-4.10.0.so
MUMPS-4.10.0-6.fc19.x86_64
MUMPS-devel-4.10.0-6.fc19.x86_64
2. Large documentation must go in a -doc subpackage.
Note: Documentation size is 1669120 bytes in 6 files.
---
I suggest not installing the .ps file, and then up to you if
creating or not a -doc package just for the .pdf. Since this
packge is expected to be a dependency of others, probably a
good idea do create a -doc package.
The ChangeLog file should go to the -devel package also.
3. SourceX / PatchY prefixed with %{name}.
Note: Patch0 (examples-mpilibs.patch) Patch1 (shared-pord.patch) Source1
(Makefile.par.inc) Patch2 (shared-mumps.patch)
---
You may rename the sources to prefix it with MUMPS-, but it is
optional (just to reduce the number the warnings)
Extra:
1. Add openssh-clients to BuildRequires, optionally making all the
%check conditional. Google'ing a bit I found the reason it would
fail by missing ssh or rsh:
http://www.open-mpi.org/community/lists/users/2010/07/13503.php
2. About the undefined-non-weak-symbol and unused-direct-shlib-dependency,
unless someone else has some comments, I believe it is ok, as the
package is functional. And it appears bogus, for example. it tells:
W: unused-direct-shlib-dependency /usr/lib64/libsmumps-4.10.0.so
libmpi_f77.so.1
and then
W: undefined-non-weak-symbol /usr/lib64/libdmumps-4.10.0.so mpi_reduce_
but the mpi_reduce_ symbol is defined in libmpi_f77.so.1
The warning should be either because of the mpi libraries not in
LD_LIBRARY_PATH at rpmlint runtime, or maybe (unlikely) because of the
reversed links in the main package and libraries in the -devel one
as explained above.
Fixed. :)
Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec
SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-7.fc18.src.rpm
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=ywxWzihhCYa=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #17 from Antonio Trande trp...@katamail.com --- (In reply to comment #16) 3. The variable OPTL appears to be the best place to use -Wl,-as-needed, that is, the pseudo patch: -OPTL= -O -OPTL= -Wl,-as-needed These flags now are inserted directly in 'shared-mumps.patch' for 'libopen-pal' and 'libopen-rte' libraries expressly indicated; this repairs the problem during installation already mentioned since first comment (#1). As well it repairs 'unused-direct-shlib-dependency' warnings already mentioned in comment#4: MUMPS.x86_64: W: unused-direct-shlib-dependency /usr/lib64/libdmumps-4.10.0.so libopen-pal.so.4 MUMPS.x86_64: W: unused-direct-shlib-dependency /usr/lib64/libsmumps-4.10.0.so libopen-pal.so.4 Furthermore, I've added '-Wuninitialized -Wno-maybe-uninitialized' in shared-mumps.patch to silence '-Wmaybe-uninizialized' warnings. Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-6.fc18.src.rpm -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=8Y3yVKcSDHa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #16 from Antonio Trande trp...@katamail.com ---
(In reply to comment #15)
You should bump the release for every review request update,
that is, the last one should be Release 4.
Yes.
Things that still should be done include:
1. You added
sed -e 's|@@CFLAGS@@|%{optflags}|g' -i Makefile.inc
to the spec as I suggested, but it is pointless, because
SOURCE1 (Makefile.seq.inc) and SOURCE2 (Makefile.par.inc)
still hardcode CFLAGS, that is, you should change those
to the pseudo patch:
-OPTC= -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions
-fstack-protector --param=ssp-buffer-size=4
+OPTC= -@@CFLAGS@@
2. Following above, the OPTF variable should also use
%{optflags}, and not the hardcoded -O -Dintel_ -DALLOW_NON_INIT
that is, should replace the -O with %{optflags}.
Done.
3. The variable OPTL appears to be the best place to use
-Wl,-as-needed, that is, the pseudo patch:
-OPTL= -O
-OPTL= -Wl,-as-needed
Okay but I don't know where impose this flags.
Long history short, for the sake of review, I would
be satisfied if you do the simpler approach of
dropping the seq build, and only do the par
build.
I agree.
I remove 'seq' sub-package building.
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=QRsMLEF3bBa=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #12 from Antonio Trande trp...@katamail.com --- Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-3.fc18.src.rpm - Following the upstream indications, I've linked openmpi header/libs files to build sequential version. This seems fix the examples' errors :D : $ mpirun -np 2 ./ssimpletest input_simpletest_real Entering SMUMPS driver with JOB, N, NZ = 6 5 12 SMUMPS 4.10.0 L U Solver for unsymmetric matrices Type of parallelism: Working host ** ANALYSIS STEP ... Structural symmetry (in percent)= 92 Density: NBdense, Average, Median =021 ... No column permutation Ordering based on AMF Leaving analysis phase with ... INFOG(1) = 0 INFOG(2) = 0 -- (20) Number of entries in factors (estim.) = 15 -- (3) Storage of factors (REAL, estimated) = 15 -- (4) Storage of factors (INT , estimated) = 50 -- (5) Maximum frontal size (estimated) = 3 -- (6) Number of nodes in the tree = 3 -- (32) Type of analysis effectively used = 1 -- (7) Ordering option effectively used = 2 ICNTL(6) Maximum transversal option= 0 ICNTL(7) Pivot order option= 7 Percentage of memory relaxation (effective)= 20 Number of level 2 nodes= 0 Number of split nodes = 0 RINFOG(1) Operations during elimination (estim)= 1.900D+01 ** Rank of proc needing largest memory in IC facto: 0 ** Estimated corresponding MBYTES for IC facto: 1 ** Estimated avg. MBYTES per work. proc at facto (IC) : 1 ** TOTAL space in MBYTES for IC factorization : 2 ** Rank of proc needing largest memory for OOC facto : 0 ** Estimated corresponding MBYTES for OOC facto : 1 ** Estimated avg. MBYTES per work. proc at facto (OOC): 1 ** TOTAL space in MBYTES for OOC factorization: 2 ** FACTORIZATION STEP GLOBAL STATISTICS PRIOR NUMERICAL FACTORIZATION ... NUMBER OF WORKING PROCESSES = 2 OUT-OF-CORE OPTION (ICNTL(22)) = 0 REAL SPACE FOR FACTORS = 15 INTEGER SPACE FOR FACTORS= 50 MAXIMUM FRONTAL SIZE (ESTIMATED) = 3 NUMBER OF NODES IN THE TREE = 3 Convergence error after scaling for ONE-NORM (option 7/8) = 0.38D+00 Maximum effective relaxed size of S = 147 Average effective relaxed size of S = 143 GLOBAL TIME FOR MATRIX DISTRIBUTION = 0. ** Memory relaxation parameter ( ICNTL(14) ):20 ** Rank of processor needing largest memory in facto : 0 ** Space in MBYTES used by this processor for facto : 1 ** Avg. Space in MBYTES per working proc during facto: 1 ELAPSED TIME FOR FACTORIZATION = 0.0008 Maximum effective space used in S (KEEP8(67) = 12 Average effective space used in S (KEEP8(67) = 8 ** EFF Min: Rank of processor needing largest memory : 0 ** EFF Min: Space in MBYTES used by this processor : 1 ** EFF Min: Avg. Space in MBYTES per working proc: 1 GLOBAL STATISTICS RINFOG(2) OPERATIONS IN NODE ASSEMBLY = 2.000D+00 --(3) OPERATIONS IN NODE ELIMINATION= 1.900D+01 INFOG (9) REAL SPACE FOR FACTORS= 15 INFOG(10) INTEGER SPACE FOR FACTORS = 50 INFOG(11) MAXIMUM FRONT SIZE= 3 INFOG(29) NUMBER OF ENTRIES IN FACTORS = 15 INFOG(12) NB OF OFF DIAGONAL PIVOTS = 0 INFOG(13) NUMBER OF DELAYED PIVOTS = 0 INFOG(14) NUMBER OF MEMORY COMPRESS = 0 KEEP8(108) Extra copies IP stacking = 0 ** SOLVE CHECK STEP STATISTICS PRIOR SOLVE PHASE ... NUMBER OF RIGHT-HAND-SIDES= 1 BLOCKING FACTOR FOR MULTIPLE RHS = 1 ICNTL (9) = 1 --- (10) = 0 --- (11) = 0 --- (20) = 0 --- (21) = 0 --- (30) = 0 ** Rank of processor needing largest memory in solve : 0 ** Space in MBYTES used by this processor
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com changed: What|Removed |Added Assignee|nob...@fedoraproject.org|paulo.cesar.pereira.de.andr ||a...@gmail.com --- Comment #13 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com --- (In reply to comment #11) (In reply to comment #10) (In reply to comment #9) At this point, I'm blocked by my knowledges which are limited in this field. You think I had to study for a long time how Makefile.inc are used during compilation and probably it would have not been so difficult for an expert programmer. I think you should focus first on getting the examples to show something more helpful :-) The error messages ERROR in MPI_ALLREDUCE, DATATYPE= 7 are not much encouraging. Hi Paulo. Hi Antonio, I have contacted maintainer team about this error: https://listes.ens-lyon.fr/sympa/arc/mumps-users/2013-02/msg00030.html This error may be tied with the fake 'mpi.h' file included in the source. I think I'm not able to take on these issues, if you want take this package, I would agree. Or in case I can co-maintain it with you. You can do it :-) I even assigned the bug to myself so I could act as your sponsor. I suggest you add a %check in the spec, somewhat like: %check module load mpi LD_LIBRARY_PATH=$PWD:../libmumps_seq:$LD_LIBRARY_PATH ./ssimpletest input_simpletest_real LD_LIBRARY_PATH=$PWD:../lib:$LD_LIBRARY_PATH mpirun -np 2 ./ssimpletest input_simpletest_real I think it should also have some proper clarification of when and how libmpiseq-4.10.0.so is used, also, it does not have a simpler libmpiseq.so symlink to it like the other libraries generated. Is it really required to generate the sequential mumps in the Fedora environment? If yes, maybe should also make a different examples build exercising it. If not, probably could simplify significantly the spec by making only the non sequential build. About the blacs_gridexit_ it is defined at: $ objdump -T /usr/lib64/openmpi/lib/libmpiblacs.so.1| grep blacs_gridexit_ 9d40 gDF .text 0103 Baseblacs_gridexit_ and libzmumps.so is linked to it, so, it looks more likely rpmlint does not resolve it because it did not run module load mpi in its environment. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=n1wg16AyJKa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #14 from Antonio Trande trp...@katamail.com --- (In reply to comment #13) You can do it :-) I even assigned the bug to myself so I could act as your sponsor. Great ! :D I suggest you add a %check in the spec, somewhat like: %check module load mpi LD_LIBRARY_PATH=$PWD:../libmumps_seq:$LD_LIBRARY_PATH ./ssimpletest input_simpletest_real LD_LIBRARY_PATH=$PWD:../lib:$LD_LIBRARY_PATH mpirun -np 2 ./ssimpletest input_simpletest_real Done. Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-3.fc18.src.rpm I think it should also have some proper clarification of when and how libmpiseq-4.10.0.so is used, also, it does not have a simpler libmpiseq.so symlink to it like the other libraries generated. Is it really required to generate the sequential mumps in the Fedora environment? If yes, maybe should also make a different examples build exercising it. If not, probably could simplify significantly the spec by making only the non sequential build. Well, this is what's reported in README file: Sequential version -- You can use the parallel MPI version of MUMPS on a single processor. If you only plan to use MUMPS on a uniprocessor machine, and do not want to install parallel libraries such as MPI, ScaLAPACK, etc... then it might be more convenient to use one of the Makefile.ARCH.SEQ to build a sequential version of MUMPS instead of a parallel one. It seems me that can be used the only parallel version. Hold the sequential one can be an option. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=3qkfrMVxEza=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #15 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com
---
You should bump the release for every review request update,
that is, the last one should be Release 4.
Things that still should be done include:
1. You added
sed -e 's|@@CFLAGS@@|%{optflags}|g' -i Makefile.inc
to the spec as I suggested, but it is pointless, because
SOURCE1 (Makefile.seq.inc) and SOURCE2 (Makefile.par.inc)
still hardcode CFLAGS, that is, you should change those
to the pseudo patch:
-OPTC= -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions
-fstack-protector --param=ssp-buffer-size=4
+OPTC= -@@CFLAGS@@
2. Following above, the OPTF variable should also use
%{optflags}, and not the hardcoded -O -Dintel_ -DALLOW_NON_INIT
that is, should replace the -O with %{optflags}.
3. The variable OPTL appears to be the best place to use
-Wl,-as-needed, that is, the pseudo patch:
-OPTL= -O
-OPTL= -Wl,-as-needed
About the sequential vs parallel versions, it may not be a
runtime choice, so, should also build a version of the examples
using it the seq libraries, and test them. Should be something
like:
cp -ar examples examples.seq
sed -e 's|libdir = $(topdir)/lib|libdir = $(topdir)/libmums_seq' -i
examples.seq/Makefile
This would also need some extra patching to add a
examples.seq target to the toplevel Makefile. Oterhwise,
you cannot know if the sequential packages and libraries
are functional. I can see the seq libraries as an
option when there is no functional mpi, otherwise, I
think it would be easier to just build the parallel
version.
Either way, I think it is doing a bogus build in the
libmumps_seq directory, because, if I understand it
correctly, libmpiseq-4.10.0.so should be a wrapper
to not need to link to the mpi libraries, but every
other library in that directory is linked to the
mpi libraries. And these chunks are duplicated
in the spec for the seq and par builds:
MUMPS_MPI=openmpi
MUMPS_INCDIR=-I/usr/include/openmpi-%{_arch}
MUMPS_LIBF77=\
-L%{_libdir}/openmpi -L%{_libdir}/openmpi/lib \
-lmpi_f77 -lmpi -lopen-rte \
-lopen-pal -lscalapack -lmpiblacs \
-lmpiblacsF77init -lmpiblacsCinit -llapack
make MUMPS_MPI=$MUMPS_MPI \
MUMPS_INCDIR=$MUMPS_INCDIR \
MUMPS_LIBF77=$MUMPS_LIBF77 \
all
Maybe the libraries should not be explicitly
linked to the mpi libraries? And only actual
binaries linked to; hmm, sanity tells it should
be other way around to allow runtime choice, so,
the MUMPS-seq package should include the libraries
in the libmumps_seq directory (and those cannot be
linked to mpi, but to libmpiseq-4.10.0.so), and then,
have the 2 packages conflict with each other, or
use alternatives or environment-modules to allow
both to be installed at the same time.
Long history short, for the sake of review, I would
be satisfied if you do the simpler approach of
dropping the seq build, and only do the par
build.
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=thaVhR2hIPa=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #11 from Antonio Trande trp...@katamail.com --- (In reply to comment #10) (In reply to comment #9) At this point, I'm blocked by my knowledges which are limited in this field. You think I had to study for a long time how Makefile.inc are used during compilation and probably it would have not been so difficult for an expert programmer. I think you should focus first on getting the examples to show something more helpful :-) The error messages ERROR in MPI_ALLREDUCE, DATATYPE= 7 are not much encouraging. Hi Paulo. I have contacted maintainer team about this error: https://listes.ens-lyon.fr/sympa/arc/mumps-users/2013-02/msg00030.html This error may be tied with the fake 'mpi.h' file included in the source. I think I'm not able to take on these issues, if you want take this package, I would agree. Or in case I can co-maintain it with you. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=M47hKfff0Ya=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #7 from Antonio Trande trp...@katamail.com ---
(In reply to comment #6)
[...]
MUMPS.x86_64: W: unused-direct-shlib-dependency
/usr/lib64/libsmumps-4.10.0.so /lib64/libquadmath.so.0
4 packages and 0 specfiles checked; 1 errors, 61 warnings.
# echo 'rpmlint-done:'
'unused-direct-shlib-dependency' warnings seem the origin of dependency
problems during installation.
Sincerely, I don't know how resolve them. After a searching, it seems that
the flags '-Wl --as-needed' can be useful but it seems me they do not work.
I am not sure about this either, using the last example, the the link command
line does not have an explicit -lquadmath, and it should not be linked by
gfotran:
$ grep quadmath /usr/lib/gcc/x86_64-redhat-linux/4.8.0/libgfortran.spec
*lib: %{static-libgfortran:--as-needed} -lquadmath
%{static-libgfortran:--no-as-needed} -lm %(libgcc) %(liborig)
but something is pulling it.
the link command line is:
gfortran -shared smumps_part1.o smumps_part2.o smumps_part3.o smumps_part4.o
smumps_part5.o smumps_part6.o smumps_part7.o smumps_part8.o
smumps_comm_buffer.o smumps_load.o smumps_c.o smumps_ooc_buffer.o
smumps_ooc.o smumps_struc_def.o -Wl,-soname,libsmumps-4.10.0.so -L../lib
-lmumps_common -L../PORD/lib/ -lpord -L../libseq -lmpiseq -lblas -o
../lib/libsmumps-4.10.0.so -Wl,-z,defs
Do you think we can't do nothing ?
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=TQ3w9qLfP0a=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #8 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com --- I think the command line is somewhat messed, and should add linker (-Wl) commands before objects, but the -Wl,soname... last in the command line. Order of options (usually) matter when linking. If adding -Wl,-as-needed, it should also come before objects, or it would not consider the object files specified after it, and you would have an underlink issue instead of overlink. Why it must be called libsmumps-4.10.0.so and not libsmumps.so.4.10.0 ? -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=u3nkxjVQiba=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #9 from Antonio Trande trp...@katamail.com --- (In reply to comment #8) I think the command line is somewhat messed, and should add linker (-Wl) commands before objects, but the -Wl,soname... last in the command line. Order of options (usually) matter when linking. If adding -Wl,-as-needed, it should also come before objects, or it would not consider the object files specified after it, and you would have an underlink issue instead of overlink. Why it must be called libsmumps-4.10.0.so and not libsmumps.so.4.10.0 ? At this point, I'm blocked by my knowledges which are limited in this field. You think I had to study for a long time how Makefile.inc are used during compilation and probably it would have not been so difficult for an expert programmer. About command line flags, I've not idea why he who has written the patches then has arranged those flags in that way. :) How would you write that line ? -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=3xf2DCfyc8a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #10 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com
---
(In reply to comment #9)
At this point, I'm blocked by my knowledges which are limited in this field.
You think I had to study for a long time how Makefile.inc are used during
compilation and probably it would have not been so difficult for an expert
programmer.
I think you should focus first on getting the examples
to show something more helpful :-) The error messages
ERROR in MPI_ALLREDUCE, DATATYPE= 7
are not much encouraging.
About command line flags, I've not idea why he who has written the patches
then has arranged those flags in that way. :)
Usually this is not much of an issue, and it it only
a rpmlint warning.
There is a small chance of problems if one intends
to run debian binaries in fedora if changing the soname
to match the package version, so, probably better to
not let it be the cause of incompatibility.
How would you write that line ?
I would experiment a bit, no ready response, but I
would at first move the -Wl,-z,defs earlier in the
command line, and ensure any -L and -l are before
object file names, and also add -Wl,-as-needed.
BTW, you should not hardcode CFLAGS in the SOURCEX
Makefile.*, and instead patch them. Could add a regex
pattern there like @@CFLAGS@@, and in the spec run
sed -e 's|@@CFLAGS@@|%{optflags}|g' -i Makefile.*
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=et5CwDqR56a=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #5 from Antonio Trande trp...@katamail.com ---
Hi Paulo.
Thank you for cooperation.
I'm dedicated myself to the Make* files insomuch as I've forgotten all that in
evidence. :P
(In reply to comment #4)
Since this package intersects with your ipopt, and my
coin-or-Ipopt review request, I did a simple review
request:
MUST:
1. %build honors applicable compiler flags or justifies otherwise.
Build does not honor compiler flags. See grep -r CFLAGS and
grep -r OPTC in the BUILD dir. Need to either patch, or just
pass CFLAGS to make call.
I've added all flags in the Make.par.inc and Make.seq.inc files directly.
2. Changelog in prescribed format.
Minor issue, just break the line between the 2 changelog entries.
Fixed.
3. Package does not run rm -rf %{buildroot} (or $RPM_BUILD_ROOT) at the
beginning of %install.
Self explanatory. Also remove the %clean section.
Fixed.
4. Each %files section contains %defattr if rpm 4.4
Note: %defattr present but not needed
Remove %defattr from all %files sections.
Fixed.
5. Development files must be in a -devel package
.so links are not in the -devel package and also should use
a relative link (target is in the same directory).
Fixed.
6. Package does not contain duplicates in %files.
Note: warning: File listed twice: /usr/lib64/MUMPS
Self explanatory.
Fixed.
7. If (and only if) the source package includes the text of the license(s)
in its own file, then that file, containing the text of the license(s)
for the package is included in %doc.
LICENSE file is not in %doc.
README is installed and not set as %doc.
Package is not owner of %{_docdir}/%{name}-%{version}.
Since the -seq package does not require the main package (error?)
it should also include LICENSE as %doc; I suggest making it require
the main package to also silence the warning about it not requiring
the main package...
The .ps and .pdf documentation is not installed. I suggest
%doc doc/*.pdf and %doc doc/%.ps
Fixed.
9. Package must own all directories that it creates.
As described in 8., package is not owner of %{_docdir}/%{name}-%{version}.
Fixed.
SHOULD:
1. Buildroot is not present
Note: Buildroot: present but not needed
Just remove the
BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
line.
Fixed.
2. Package has no %clean section with rm -rf %{buildroot} (or
$RPM_BUILD_ROOT)
Note: %clean present but not required
Just remove the %clean section.
Fixed.
3. Spec use %global instead of %define.
Note: %define __provides_exclude_from
%{_libdir}/openmpi/lib/(lib(mca|o|v)|openmpi/).*.so %define
__requires_exclude libopen.*.so
Self explanatory.
Fixed.
EXTRA:
1. Please consider installing headers in /usr/include/MUMPS:
$ ls rpms-unpacked/MUMPS-devel-4.10.0-1.fc18.x86_64.rpm/usr/include/|wc -l
14
this may need some minor patching or adding -IMUMPS to build.
Fixed.
2. -devel package should have:
Requires: %{name}%{?_isa} = %{version}-%{release}
instead of
Requires: %{name} = %{version}-%{release}
Fixed.
3. Summary and description of the seq subpackage are a wrong, cutpaste
of the -devel package.
Fixed.
4. Please make a examples subpackage with the examples subdir, verify
the README there, and validate it works. *OR* make of the examples
subdirectory a %check section. This would help a reviewer to press
x in the Package functions as described field.
I've added a 'examples' sub-package that contains README and all 'simpletest'
files. But it seems to exist another problem by executing them:
$ ./ssimpletest input_simpletest_real
./ssimpletest: error while loading shared libraries: libmpi_f77.so.1: cannot
open shared object file: No such file or directory
5. You should not do
chmod a-x $RPM_BUILD_ROOT%{_libdir}/%{name}/?simpletest
chmod a-x $RPM_BUILD_ROOT%{_libdir}/%{name}/input_*
in %install. This just causes it to install ELF binary files without
executable permission just for the bogus warning
'script-without-shebang'
Fixed.
6. Instead of using a single mumps-extra.tar.gz tarball, You should add
it as a Patch, with the proper link to
http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/raring/mumps/raring/
files/head:/debian/patches/$PATCH_NAME
and use %patch instead of calling patch in %build. At least, should
use patch in %prep (that is, to help in without prep partial
debug builds).
Should also give a brief description of what the patch does, and
if applicable, a link to where the patch was submitted upstream.
Fixed.
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
--- Comment #6 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com ---
4. Please make a examples subpackage with the examples subdir, verify
the README there, and validate it works. *OR* make of the examples
subdirectory a %check section. This would help a reviewer to press
x in the Package functions as described field.
I've added a 'examples' sub-package that contains README and all
'simpletest' files. But it seems to exist another problem by executing them:
$ ./ssimpletest input_simpletest_real
./ssimpletest: error while loading shared libraries: libmpi_f77.so.1: cannot
open shared object file: No such file or directory
You probably need write some wrapper script(s) to
module load mpi
$ cd BUILD/MUMPS_4.10.0/examples/
$ ./ssimpletest input_simpletest_real
./ssimpletest: error while loading shared libraries: libmpi_f77.so.1: cannot
open shared object file: No such file or directory
$ module load mpi
$ ./ssimpletest input_simpletest_real
ERROR in MPI_ALLREDUCE, DATATYPE= 7
$ ./dsimpletest input_simpletest_real
ERROR in MPI_ALLREDUCE, DATATYPE= 7
$ mpirun -np 2 ./ssimpletest input_simpletest_real
ERROR in MPI_ALLREDUCE, DATATYPE= 7
ERROR in MPI_ALLREDUCE, DATATYPE= 7
It appears to not be working...
[...]
MUMPS.x86_64: W: unused-direct-shlib-dependency
/usr/lib64/libsmumps-4.10.0.so /lib64/libquadmath.so.0
4 packages and 0 specfiles checked; 1 errors, 61 warnings.
# echo 'rpmlint-done:'
'unused-direct-shlib-dependency' warnings seem the origin of dependency
problems during installation.
Sincerely, I don't know how resolve them. After a searching, it seems that
the flags '-Wl --as-needed' can be useful but it seems me they do not work.
I am not sure about this either, using the last example, the the link command
line does not have an explicit -lquadmath, and it should not be linked by
gfotran:
$ grep quadmath /usr/lib/gcc/x86_64-redhat-linux/4.8.0/libgfortran.spec
*lib: %{static-libgfortran:--as-needed} -lquadmath
%{static-libgfortran:--no-as-needed} -lm %(libgcc) %(liborig)
but something is pulling it.
the link command line is:
gfortran -shared smumps_part1.o smumps_part2.o smumps_part3.o smumps_part4.o
smumps_part5.o smumps_part6.o smumps_part7.o smumps_part8.o
smumps_comm_buffer.o smumps_load.o smumps_c.o smumps_ooc_buffer.o smumps_ooc.o
smumps_struc_def.o -Wl,-soname,libsmumps-4.10.0.so -L../lib -lmumps_common
-L../PORD/lib/ -lpord -L../libseq -lmpiseq -lblas -o
../lib/libsmumps-4.10.0.so -Wl,-z,defs
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=J90jSPO6F2a=cc_unsubscribe
___
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #2 from Antonio Trande trp...@katamail.com --- (In reply to comment #1) Run repoquery --whatprovides 'libopen-rte.4()(64bit)' to find that nothing provides that library SONAME. Hence nothing satisfies the dependency. The library file is include in some non-public path in openmpi-1.6.3-7.fc18.x86_64.rpm - http://koji.fedoraproject.org/koji/rpminfo?rpmID=3664682 - but it seems the automatic Provides are filtered. The %changelog mentions bug 741104 (Exclude private libraries from provides/requires). Some investigation is needed. Hi Michael. I have added %__requires_exclude_* macros to prevent the libraries request from openmpi. openmpi downgrading seems not resolve the problem. Spec URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS.spec SRPM URL: http://sagitter.fedorapeople.org/MUMPS/MUMPS-4.10.0-1.fc18.src.rpm -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=OOeQIdZvE6a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Michael Schwendt mschwe...@gmail.com changed: What|Removed |Added CC||mschwe...@gmail.com --- Comment #3 from Michael Schwendt mschwe...@gmail.com --- That sounds questionable. Either the libs are _private_ and nothing external ought to depend on them - then it's okay to hide their automatic Provides. Or external pieces may depend on them, and then it would not be okay to kill the RPM dependencies in any of the packages. I'll mail openmpi-owner(s) and Cc you. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=plCOIQkp4ya=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=913152
Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com changed:
What|Removed |Added
CC||paulo.cesar.pereira.de.andr
||a...@gmail.com
--- Comment #4 from Paulo Andrade paulo.cesar.pereira.de.andr...@gmail.com ---
Since this package intersects with your ipopt, and my
coin-or-Ipopt review request, I did a simple review
request:
MUST:
1. %build honors applicable compiler flags or justifies otherwise.
Build does not honor compiler flags. See grep -r CFLAGS and
grep -r OPTC in the BUILD dir. Need to either patch, or just
pass CFLAGS to make call.
2. Changelog in prescribed format.
Minor issue, just break the line between the 2 changelog entries.
3. Package does not run rm -rf %{buildroot} (or $RPM_BUILD_ROOT) at the
beginning of %install.
Self explanatory. Also remove the %clean section.
4. Each %files section contains %defattr if rpm 4.4
Note: %defattr present but not needed
Remove %defattr from all %files sections.
5. Development files must be in a -devel package
.so links are not in the -devel package and also should use
a relative link (target is in the same directory).
6. Package does not contain duplicates in %files.
Note: warning: File listed twice: /usr/lib64/MUMPS
Self explanatory.
7. If (and only if) the source package includes the text of the license(s)
in its own file, then that file, containing the text of the license(s)
for the package is included in %doc.
LICENSE file is not in %doc.
README is installed and not set as %doc.
Package is not owner of %{_docdir}/%{name}-%{version}.
Since the -seq package does not require the main package (error?)
it should also include LICENSE as %doc; I suggest making it require
the main package to also silence the warning about it not requiring
the main package...
The .ps and .pdf documentation is not installed. I suggest
%doc doc/*.pdf and %doc doc/%.ps
8. License file installed when any subpackage combination is installed.
See 7.
9. Package must own all directories that it creates.
As described in 8., package is not owner of %{_docdir}/%{name}-%{version}.
SHOULD:
1. Buildroot is not present
Note: Buildroot: present but not needed
Just remove the
BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
line.
2. Package has no %clean section with rm -rf %{buildroot} (or
$RPM_BUILD_ROOT)
Note: %clean present but not required
Just remove the %clean section.
3. Spec use %global instead of %define.
Note: %define __provides_exclude_from
%{_libdir}/openmpi/lib/(lib(mca|o|v)|openmpi/).*.so %define
__requires_exclude libopen.*.so
Self explanatory.
EXTRA:
1. Please consider installing headers in /usr/include/MUMPS:
$ ls rpms-unpacked/MUMPS-devel-4.10.0-1.fc18.x86_64.rpm/usr/include/|wc -l
14
this may need some minor patching or adding -IMUMPS to build.
2. -devel package should have:
Requires: %{name}%{?_isa} = %{version}-%{release}
instead of
Requires: %{name} = %{version}-%{release}
3. Summary and description of the seq subpackage are a wrong, cutpaste
of the -devel package.
4. Please make a examples subpackage with the examples subdir, verify
the README there, and validate it works. *OR* make of the examples
subdirectory a %check section. This would help a reviewer to press
x in the Package functions as described field.
5. You should not do
chmod a-x $RPM_BUILD_ROOT%{_libdir}/%{name}/?simpletest
chmod a-x $RPM_BUILD_ROOT%{_libdir}/%{name}/input_*
in %install. This just causes it to install ELF binary files without
executable permission just for the bogus warning
'script-without-shebang'
6. Instead of using a single mumps-extra.tar.gz tarball, You should add
it as a Patch, with the proper link to
http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/raring/mumps/raring/files/head:/debian/patches/$PATCH_NAME
and use %patch instead of calling patch in %build. At least, should
use patch in %prep (that is, to help in without prep partial
debug builds).
Should also give a brief description of what the patch does, and
if applicable, a link to where the patch was submitted upstream.
==
Package Review
==
Key:
[x] = Pass
[!] = Fail
[-] = Not applicable
[?] = Not evaluated
[ ] = Manual review needed
Issues:
===
[!]: Development (unversioned) .so files in -devel subpackage, if present.
Note: Unversioned so-files directly in %_libdir.
See: http://fedoraproject.org/wiki/Packaging/Guidelines#DevelPackages
[!]:
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Antonio Trande trp...@katamail.com changed: What|Removed |Added Clone Of||566750 CC||trp...@katamail.com Blocks||177841 (FE-NEEDSPONSOR) -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=vWxZnOkhoCa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 Antonio Trande trp...@katamail.com changed: What|Removed |Added Depends On||908089 -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=ejoLOKPg98a=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
[Bug 913152] Review Request: MUMPS - A MUltifrontal Massively Parallel sparse direct Solver
Product: Fedora https://bugzilla.redhat.com/show_bug.cgi?id=913152 --- Comment #1 from Michael Schwendt mschwe...@gmail.com --- Run repoquery --whatprovides 'libopen-rte.4()(64bit)' to find that nothing provides that library SONAME. Hence nothing satisfies the dependency. The library file is include in some non-public path in openmpi-1.6.3-7.fc18.x86_64.rpm - http://koji.fedoraproject.org/koji/rpminfo?rpmID=3664682 - but it seems the automatic Provides are filtered. The %changelog mentions bug 741104 (Exclude private libraries from provides/requires). Some investigation is needed. -- You are receiving this mail because: You are on the CC list for the bug. Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=ztwqxo0SoSa=cc_unsubscribe ___ package-review mailing list package-review@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/package-review
