Spec URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r01/quantum-espresso.spec SRPM URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r01/quantum-espresso-5.2.1-1.fc23.src.rpm Description: QUANTUM ESPRESSO is an integrated suite of Open-Source computer codes for electronic-structure calculations and materials modeling at the nanoscale. It is based on density-functional theory, plane waves, and pseudopotentials. Fedora Account System Username: marcindulak
marcindulak's scratch build of quantum-espresso-5.2.1-1.fc23.src.rpm for rawhide completed http://koji.fedoraproject.org/koji/taskinfo?taskID=11884527
Hm, what does this do?: %if 0%{?el6} %global mpich mpich %global mpich_load %_mpich_load %global mpich_unload %_mpich_unload %else %global mpich mpich %global mpich_load %_mpich_load %global mpich_unload %_mpich_unload %endif Requires are normally written with one =. I think == might be misinterpreted. You can replace the calls to `basename ${f}` with ${f%/*}, makes things slightly shorter and more readable. Please use pushd/popd instead of cd/cd -. This is recommended because the path is printed, which makes logs easier to understand. %defattr(-,root,root,-) is not needed. Use %{!?_licensedir:%global license %doc} and then you don't need conditionals in %files. Can you comment on the reason for multiple source tarballs? Are they always released together?
marcindulak's scratch build of quantum-espresso-5.2.1-1.el6.src.rpm for el6-candidate failed http://koji.fedoraproject.org/koji/taskinfo?taskID=12125885
As you can see I'm working on this. Do you know why my scratch builds get listed here? In https://kojipkgs.fedoraproject.org//work/tasks/5886/12125886/build.log you can see that setting global license does not work for me on el6 when defining this global at top of the spec. Apparently License gets confused with license.
marcindulak's scratch build of quantum-espresso-5.2.1-1.el6.src.rpm for el6-candidate completed http://koji.fedoraproject.org/koji/taskinfo?taskID=12126535
(In reply to marcindulak from comment #4) > As you can see I'm working on this. Do you know why my scratch builds get > listed here? It's a new feature based on fedmsg. Scratch builds are automatically listed on the bug page when there's a review open. > In https://kojipkgs.fedoraproject.org//work/tasks/5886/12125886/build.log > you can see that setting global license does not work for me on el6 when > defining this global at top of the spec. Apparently License gets confused > with license. I don't see the spec file you used, but here's an example that seems to work. I just did something similar today, to have the same spec file for el6, epel7, and various fedora versions: http://pkgs.fedoraproject.org/cgit/python-dialog.git/commit/?id=db10aeb59cccc1a201554719ca64b794452d2e4f.
marcindulak's scratch build of quantum-espresso-5.2.1-1.el6.src.rpm for el6-candidate completed http://koji.fedoraproject.org/koji/taskinfo?taskID=12215112
(In reply to Zbigniew Jędrzejewski-Szmek from comment #2) > Hm, what does this do?: > > %if 0%{?el6} > %global mpich mpich > %global mpich_load %_mpich_load > %global mpich_unload %_mpich_unload > %else > %global mpich mpich > %global mpich_load %_mpich_load > %global mpich_unload %_mpich_unload > %endif > this is a legacy part when there was mpich2 on el6 and mpich on Fedora. I removed that global variables. > > Requires are normally written with one =. I think == might be misinterpreted. OK > > You can replace the calls to `basename ${f}` with ${f%/*}, makes things > slightly shorter and more readable. let's keep the simple basename. I'm not a fan of shortcuts. > > Please use pushd/popd instead of cd/cd -. This is recommended because the > path is printed, which makes logs easier to understand. > OK > %defattr(-,root,root,-) is not needed. i believe fedora-review complains if %defattr(-,root,root,-) is missing > > Use %{!?_licensedir:%global license %doc} and then you don't need > conditionals in %files. > this works for me when defining this global *after* the License field. Apparently License is confused with license if global defined on the top of the spec: http://koji.fedoraproject.org/koji/getfile?taskID=12125886&name=build.log&offset=-4000 > Can you comment on the reason for multiple source tarballs? Are they always > released together? I would like we clarify two things: 1. the source includes L. Peter Deutsch md5 implementation. I think gromacs RPM also includes this md5 (src/external/tng_io/src/lib/md5.c) so we are fine? 2. quantum-espresso bundles a small part (a couple of files) belonging to http://www.tddft.org/programs/octopus/wiki/index.php/Libxc This is described in this commit message: http://qe-forge.org/gf/project/q-e/scmsvn/?action=ScmCommitDetail&scm_commit_id=16950 I have made an additional change: set ExclusiveArch due to openblas-devel. Spec URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r02/quantum-espresso.spec SRPM URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r02/quantum-espresso-5.2.1-1.el6.src.rpm
(In reply to marcindulak from comment #8) > (In reply to Zbigniew Jędrzejewski-Szmek from comment #2) > > Hm, what does this do?: > > > > %if 0%{?el6} > > %global mpich mpich > > %global mpich_load %_mpich_load > > %global mpich_unload %_mpich_unload > > %else > > %global mpich mpich > > %global mpich_load %_mpich_load > > %global mpich_unload %_mpich_unload > > %endif > > > > this is a legacy part when there was mpich2 on el6 and mpich on Fedora. > I removed that global variables. > > > > > Requires are normally written with one =. I think == might be misinterpreted. > > OK > > > > > You can replace the calls to `basename ${f}` with ${f%/*}, makes things > > slightly shorter and more readable. > > let's keep the simple basename. I'm not a fan of shortcuts. > > > > > Please use pushd/popd instead of cd/cd -. This is recommended because the > > path is printed, which makes logs easier to understand. > > > > OK > > > %defattr(-,root,root,-) is not needed. > > i believe fedora-review complains if %defattr(-,root,root,-) is missing Quite the opposite ;) Anyway the authoritative source is https://fedoraproject.org/wiki/Packaging:Guidelines#File_Permissions, which says that they are not required. > > Use %{!?_licensedir:%global license %doc} and then you don't need > > conditionals in %files. > > > > this works for me when defining this global *after* the License field. > Apparently License is confused with license if global defined on the top of > the spec: > http://koji.fedoraproject.org/koji/getfile?taskID=12125886&name=build. > log&offset=-4000 Right. > > Can you comment on the reason for multiple source tarballs? Are they always > > released together? > > I would like we clarify two things: > > 1. the source includes L. Peter Deutsch md5 implementation. I think gromacs > RPM > also includes this md5 (src/external/tng_io/src/lib/md5.c) so we are fine? > > 2. quantum-espresso bundles a small part (a couple of files) belonging to > http://www.tddft.org/programs/octopus/wiki/index.php/Libxc > This is described in this commit message: > http://qe-forge.org/gf/project/q-e/scmsvn/ > ?action=ScmCommitDetail&scm_commit_id=16950 > > I have made an additional change: set ExclusiveArch due to openblas-devel. > > Spec URL: > https://marcindulak.fedorapeople.org/packages/quantum-espresso/r02/quantum- > espresso.spec > SRPM URL: > https://marcindulak.fedorapeople.org/packages/quantum-espresso/r02/quantum- > espresso-5.2.1-1.el6.src.rpm I'll try to review this later today.
Created attachment 1107142 [details] fix requires I looked at this, as I had an earlier version in copr (not suitable for review). The questions might well not be appropriate, but I think it would help to have notes in the spec. * It won't install because it requires an arch-specific -common package; patch attached. * I'd have thought iotk should be unbundled, but I don't know if it's of more general use. * Shouldn't this build against atlas or lapack/blas on non-x86? (ppc64le seems a plausible architecture to run it on.) I know that makes it particularly suffer the BLAS mess in Fedora, but scalapack links against reference blas anyhow. * It's not using the default compilation and linking flags (not that I agree with that requirement for computational programs). * Will smp make not work? There's no comment, and the build takes a while. * Why not use elpa? (I haven't tried with this version and the Fedora elpa.) * Shouldn't the doc be installed? * Would the GUI be useful? (I don't know.)
Thanks for tackling this. It seems to be a really complicated package and I can see that a lot of work has gone into putting it in a decent shape. Some issues: openblas is ExclusiveArch x86_64 %{ix86} armv7hl ppc64le, you should probably match that. License tag is wrong. The License tag specifies the *effective* license of the *binary* package [1, 2]. So anything that is not part of the binary rpm (like install/install-sh) doesn't matter at all. If you combine GPLv2+ with BSD or LGPL or MIT, the effective license is GPLv2+. So, if a file in the binary rpm has at least one GPL licensed file, that file is GPL. If a file only had sources with more permissive licenses, than that file would have some other license. So you need to determine the effective license of all files in the binary rpm and put the result in License. I'd guess that the result is going to be License:GPLv2+. [1] https://fedoraproject.org/wiki/Packaging:LicensingGuidelines#License:_field [2] https://fedoraproject.org/wiki/Licensing:FAQ?rd=Licensing/FAQ#What_is_.22effective_license.22_and_do_I_need_to_know_that_for_the_License:_tag.3F Requires:openmpi and Requires:mpich in the q-e-openmpi and q-e-mpich look wrong to me. Normally the automatically generated dependency would be enough. E.g. q-e-mpich.i686 requires libmpi.so.12(mpich-i386) libmpifort.so.12(mpich-i386) and this should be enough. OTOH, q-e-mpich-devel should probably R: mpich-devel, and q-e-openmpi-devel should probably R: openmpi-devel. Actually, I'd suggest moving all the BuildRequires to the top. Right now it is hard to get an overview of what will be installed in the build root, and some BR are repeated. Why export OMP_NUM_THREADS=1 in %check? Maybe you could use '%{lua: for i=20,41 do print("%{SOURCE" .. i .. "} ") end}' instead of listing all the SOURCExx explicitly in cp? The comments from Dave are also all relevant: (In reply to Dave Love from comment #10) > Created attachment 1107142 [details] > fix requires This looks correct. > * It won't install because it requires an arch-specific -common package; > patch attached. Right. > * I'd have thought iotk should be unbundled, but I don't know if it's of > more general use. Hm, good question. Is it used anywhere else? > * Shouldn't this build against atlas or lapack/blas on non-x86? (ppc64le > seems a plausible architecture to run it on.) I know that makes it > particularly suffer the BLAS mess in Fedora, but scalapack links against > reference blas anyhow. openblas is available on ppc64le. So maybe building everywhere that openblas-devel is avaialable would be enough... > * It's not using the default compilation and linking flags (not that I agree > with that requirement for computational programs). > > * Will smp make not work? There's no comment, and the build takes a while. Yeah, the build takes forever. More threads would be great. > * Why not use elpa? (I haven't tried with this version and the Fedora elpa.) > > * Shouldn't the doc be installed? > > * Would the GUI be useful? (I don't know.)
Note that automatic MPI requirement generation is only available on Fedora 23+.
Can I close my eyes and pretend that F22 is already EOL? ;) But even in F22, mpich has libmpi.so.12()(64bit), and openmpi has libmpi.so.1()(64bit), so the right dependency should be installed anyway.
(In reply to Zbigniew Jędrzejewski-Szmek from comment #13) > Can I close my eyes and pretend that F22 is already EOL? ;) > > But even in F22, mpich has libmpi.so.12()(64bit), and openmpi has > libmpi.so.1()(64bit), so the right dependency should be installed anyway. What about el6, el7? I get the following on CentOS 6: $ yum provides "libmpi.so.1" compat-openmpi-1.4.3-1.2.el6.i686 : Open Message Passing Interface Repo : base Matched from: Other : libmpi.so.1 openmpi-1.8.1-1.el6.i686 : Open Message Passing Interface Repo : base Matched from: Other : libmpi.so.1
Then in epel maybe you need to add an explicit dependency on openmpi. But in Fedora is should not be necessary, even in F22.
marcindulak's scratch build of quantum-espresso-5.2.1-3.el6.src.rpm for epel7 failed http://koji.fedoraproject.org/koji/taskinfo?taskID=12243819
marcindulak's scratch build of quantum-espresso-5.2.1-3.el6.src.rpm for epel7 failed http://koji.fedoraproject.org/koji/taskinfo?taskID=12244038
marcindulak's scratch build of quantum-espresso-5.2.1-3.el6.src.rpm for epel7 completed http://koji.fedoraproject.org/koji/taskinfo?taskID=12244545
(In reply to Zbigniew Jędrzejewski-Szmek from comment #11) > Thanks for tackling this. It seems to be a really complicated package and I > can see that a lot of work has gone into putting it in a decent shape. > > Some issues: > > openblas is ExclusiveArch x86_64 %{ix86} armv7hl ppc64le, you should > probably match that. fixed > > License tag is wrong. The License tag specifies the *effective* license of > the *binary* package [1, 2]. So anything that is not part of the binary rpm > (like install/install-sh) doesn't matter at all. If you combine GPLv2+ with > BSD or LGPL or MIT, the effective license is GPLv2+. So, if a file in the > binary rpm has at least one GPL licensed file, that file is GPL. If a file > only had sources with more permissive licenses, than that file would have > some other license. So you need to determine the effective license of all > files in the binary rpm and put the result in License. I'd guess that the > result is going to be License:GPLv2+. > fixed > [1] > https://fedoraproject.org/wiki/Packaging:LicensingGuidelines#License:_field > [2] > https://fedoraproject.org/wiki/Licensing:FAQ?rd=Licensing/FAQ#What_is_. > 22effective_license.22_and_do_I_need_to_know_that_for_the_License:_tag.3F > > Requires:openmpi and Requires:mpich in the q-e-openmpi and q-e-mpich look > wrong to me. Normally the automatically generated dependency would be > enough. E.g. q-e-mpich.i686 requires > libmpi.so.12(mpich-i386) > libmpifort.so.12(mpich-i386) > and this should be enough. The builds need to work also on el6/el7. I keep those Requires for now. > > OTOH, q-e-mpich-devel should probably R: mpich-devel, and q-e-openmpi-devel > should probably R: openmpi-devel. The BuildRquires and Requires are correct for q-e-*-devel subpackages with the current packaging of openmpi and mpich. They *Requires need to be mpich and openmpi due to the paths to directories containing fortran mod files (%{_fmoddir}/openmpi%{?_cc_name_suffix}/) being owned by mpich and openmpi. > > Actually, I'd suggest moving all the BuildRequires to the top. Right now it > is hard to get an overview of what will be installed in the build root, and > some BR are repeated. I'll get more confused by moving the dependencies to one place. > > Why export OMP_NUM_THREADS=1 in %check? removed now. > > Maybe you could use '%{lua: for i=20,41 do print("%{SOURCE" .. i .. "} ") > end}' > instead of listing all the SOURCExx explicitly in cp? how can this be done? One the command line the following looks OK: $ rpm --eval '%{lua:for i=20,41 do print(string.format("cp -p %%{SOURCE%u} pseudo\n", i)) end}' but with rpmbuild I get additional quotes: ... + cp -p '%{SOURCE20}' pseudo cp: cannot stat `%{SOURCE20}': No such file or directory ... > > The comments from Dave are also all relevant: > (In reply to Dave Love from comment #10) > > Created attachment 1107142 [details] > > fix requires > This looks correct. fixed > > > * It won't install because it requires an arch-specific -common package; > > patch attached. > Right. > > > * I'd have thought iotk should be unbundled, but I don't know if it's of > > more general use. > Hm, good question. Is it used anywhere else? iotk is required by http://www.yambo-code.org/ Unbundling iotk would require some work, and I'm not willing to do this without the support from quantum-espresso developers. Yambo is currently hosted on qe-forge http://qe-forge.org/gf/project/yambo/frs/?action=FrsReleaseBrowse&frs_package_id=40 which means that it's probably related to quantum-espresso and when packaging yambo one could just BuildRequires: quantum-espresso-{openmpi,static}-devel and quantum-espresso-{openmpi,static}-static > > > * Shouldn't this build against atlas or lapack/blas on non-x86? (ppc64le > > seems a plausible architecture to run it on.) I know that makes it > > particularly suffer the BLAS mess in Fedora, but scalapack links against > > reference blas anyhow. > openblas is available on ppc64le. So maybe building everywhere that > openblas-devel > is avaialable would be enough... I prefer openblas. Had problems with atlas in the past with various codes, and openblas just worked. > > > * It's not using the default compilation and linking flags (not that I agree > > with that requirement for computational programs). %{optflags} are used now > > > > * Will smp make not work? There's no comment, and the build takes a while. > Yeah, the build takes forever. More threads would be great. make %{?_smp_mflags} failed for me > > > * Why not use elpa? (I haven't tried with this version and the Fedora elpa.) I tried to use Fedora's ELPA but several makefiles have hardcoded ELPA. I didn't want to spend my time on this and decided to simply remove the bundled ELPA. > > > > * Shouldn't the doc be installed? I believe nobody reads the docs installed on the system nowadays. > > > > * Would the GUI be useful? (I don't know.) Let's leave it for later. pwgui looks to me as a separate package. Spec URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r03/quantum-espresso.spec SRPM URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r03/quantum-espresso-5.2.1-3.el6.src.rpm
warning: bogus date in %changelog: Wed Dec 19 2015 Marcin Dulak <Marcin.Dulak> 5.1.2-3 quantum-espresso.x86_64: W: incoherent-version-in-changelog 5.1.2-3 ['5.2.1-3.fc24', '5.2.1-3'] > > > * Will smp make not work? There's no comment, and the build takes a while. > > Yeah, the build takes forever. More threads would be great. > make %{?_smp_mflags} failed for me Please al least add a comment in the spec file. It would be great to work with upstream to fix parallel compilation, and to be able to run the tests in parallel. > > > * Shouldn't the doc be installed? > I believe nobody reads the docs installed on the system nowadays. I could dispute that. Packaged docs have advantages: they work offline, they don't get out of sync with the package, they are still there if the upstream goes away. But they are not mandatory, so if you don't want to package them that's OK. > + cp -p '%{SOURCE20}' pseudo > cp: cannot stat `%{SOURCE20}': No such file or directory It needs one more expansion level: cp -p %{expand: %{lua: for i=20,41 do print("%{SOURCE"..i.."} ") end}} pseudo/ When running the build, a number of errors appear like this: Checking uspp...[warn] Epoll ADD(4) on fd 1 failed. Old events were 0; read change was 0 (none); write change was 1 (add): Operation not permitted This might be caused by trying to add /dev/null. Cf. the following python transcript: >>> import select >>> p = select.epoll() >>> f = open('/dev/null', 'r') >>> p.register(f.fileno()) Traceback (most recent call last): File "<stdin>", line 1, in <module> PermissionError: [Errno 1] Operation not permitted It fails with the same error code. Maybe the warning in tests is harmless, maybe not, please investigate that. rpmlint: quantum-espresso.x86_64: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso.x86_64: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials False positives. quantum-espresso.x86_64: W: no-documentation quantum-espresso.x86_64: W: no-manual-page-for-binary ... OK. quantum-espresso-debuginfo.x86_64: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/... Since this is only in the debug package, it's OK. You created the q-e-common subpackage. It has only one file, which is a few kilobytes. I think it would be totally fine to package this file in the main package, the savings are not worth the overhead of having another subpackage. >> > * I'd have thought iotk should be unbundled, but I don't know if it's of >> > more general use. >> Hm, good question. Is it used anywhere else? > >iotk is required by http://www.yambo-code.org/ >Unbundling iotk would require some work, and I'm not willing to do this >without the support from quantum-espresso developers. >Yambo is currently hosted on qe-forge >http://qe-forge.org/gf/project/yambo/frs/?action=FrsReleaseBrowse&frs_package_id=40 >which means that it's probably related to quantum-espresso and when packaging yambo one could just BuildRequires: >quantum-espresso-{openmpi,static}-devel and quantum-espresso-{openmpi,static}-static You should build iotk as a shared library. This is very strongly encouraged [https://fedoraproject.org/wiki/Packaging:Guidelines#Packaging_Static_Libraries]. IS there a good reason to build iotk as a static lib?
Created attachment 1107852 [details] patch to disable some tedious warnings
Created attachment 1107853 [details] lua magic
(In reply to Zbigniew Jędrzejewski-Szmek from comment #20) > > > > * Shouldn't the doc be installed? > > I believe nobody reads the docs installed on the system nowadays. > I could dispute that. Packaged docs have advantages: they work offline, they > don't get out of sync with the package, they are still there if the upstream > goes away. Right. As a user of the package, I don't want to have to install the doc separately. Examples are particularly valuable. > But they are not mandatory, so if you don't want to package them > that's OK. It is a "should" item in the guidelines, and trivial to do, at least in this case. It's unfortunate if Fedora loses documentation, and I'd definitely rather have it than full regression tests. > When running the build, a number of errors appear like this: > Checking uspp...[warn] Epoll ADD(4) on fd 1 failed. Old events were 0; read > change was 0 (none); write change was 1 (add): Operation not permitted I don't see that, but there are several "discrepancies" listed when it's run here under RHEL6 on sandybridge. I doubt the numerical differences are important, but some differences in iteration counts look surprising. However, I'd meant to say that I don't think it's worthwhile to run complete sets of regression tests in package builds, especially when they take so long and require the extra data. One or two sanity checks from the examples directory seem enough. (The increased build resources are relevant to installers if you need to rebuild for a different MPI, as I do.) I think the MPI sub-packages should have proper Summary contents, particularly as the main one may not be installed; "%{name} - openmpi version" isn't useful, e.g. in the output of "rpm -qa". Probably there should be a guideline for that.
(In reply to Dave Love from comment #23) > (In reply to Zbigniew Jędrzejewski-Szmek from comment #20) > > When running the build, a number of errors appear like this: > > Checking uspp...[warn] Epoll ADD(4) on fd 1 failed. Old events were 0; read > > change was 0 (none); write change was 1 (add): Operation not permitted > > I don't see that, but there are several "discrepancies" listed when it's > run here under RHEL6 on sandybridge. I doubt the numerical differences > are important, but some differences in iteration counts look surprising. > > However, I'd meant to say that I don't think it's worthwhile to run > complete sets of regression tests in package builds, especially when > they take so long and require the extra data. One or two sanity > checks from the examples directory seem enough. (The increased build > resources are relevant to installers if you need to rebuild for a > different MPI, as I do.) Such tests sometimes expose compilation or memory issues or endianness errors... Scientific packages like quantum espresso don't have many users, so it is possible that on less popular architectures almost nobody will test the package. Having a big test suite that really exercises the code in realistic way is my experience very valuable. > I think the MPI sub-packages should have proper Summary contents, > particularly as the main one may not be installed; > "%{name} - openmpi version" isn't useful, e.g. in the output of > "rpm -qa". Probably there should be a guideline for that. Yeah, I also wanted to comment on this when doing the last review, but in the end I didn't have an idea that would be significantly better. What actual replacement would you propose? Summary: A suite for electronic-structure calculations and materials modeling Summary: %{name} - devel Summary: %{name} - static libraries Summary: %{name} - openmpi version Summary: %{name} - devel openmpi version Summary: %{name} - static libraries openmpi version Summary: %{name} - mpich version Summary: %{name} - devel mpich version Summary: %{name} - static libraries mpich version Summary: %{name} - common files
Thanks for patches. I have made the following changes: - fixed incorrect changelog version - disabled compilation warnings - used lua for copying pseudos - removed common package Answers to other questions: - I don't build shared iotk because the static one is referred to in the Makefiles. I don't want to patch the Makefiles for iotk at each espresso release. - "Epoll ADD(4) on fd 1 failed..." i think it's a warning coming from openmpi. I have seen openmpi warnings come and go with different openmpi versions or choices of build/run parameters. Have no time for investigate this further. Spec URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r04/quantum-espresso.spec SRPM URL: https://marcindulak.fedorapeople.org/packages/quantum-espresso/r04/quantum-espresso-5.2.1-4.el6.src.rpm
marcindulak's scratch build of quantum-espresso-5.2.1-4.el6.src.rpm for epel7 completed http://koji.fedoraproject.org/koji/taskinfo?taskID=12449155
(In reply to marcindulak from comment #25) > Answers to other questions: > - I don't build shared iotk because the static one is referred to in the > Makefiles. I don't want to patch the Makefiles for iotk at each espresso > release. OK. I think it would still be valuable to build the shared library, but it would mostly matter for packages linking to libiotk.a. If such packages are packaged for Fedora in the future, this might be worth revisiting. > - "Epoll ADD(4) on fd 1 failed..." i think it's a warning coming from > openmpi. I have seen openmpi warnings come and go with different openmpi > versions or choices of build/run parameters. Have no time for investigate > this further. OK. I seems that the build is not parallelized... Would be nice to fix that. rpmlint: quantum-espresso.src:178: W: mixed-use-of-spaces-and-tabs (spaces: line 178, tab: line 5) quantum-espresso.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso.i686: W: spelling-error %description -l en_US executables -> executable, executable s, executrices quantum-espresso.i686: W: no-documentation quantum-espresso.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso/License quantum-espresso.i686: W: no-manual-page-for-binary fd_ifc.x quantum-espresso.i686: W: no-manual-page-for-binary pw4gww.x quantum-espresso.i686: W: no-manual-page-for-binary plotrho.x quantum-espresso.i686: W: no-manual-page-for-binary fd_ef.x quantum-espresso.i686: W: no-manual-page-for-binary cppp.x quantum-espresso.i686: W: no-manual-page-for-binary wannier_ham.x quantum-espresso.i686: W: no-manual-page-for-binary path_interpolation.x quantum-espresso.i686: W: no-manual-page-for-binary generate_rVV10_kernel_table.x quantum-espresso.i686: W: no-manual-page-for-binary molecularpdos.x quantum-espresso.i686: W: no-manual-page-for-binary pp.x quantum-espresso.i686: W: no-manual-page-for-binary pw2wannier90.x quantum-espresso.i686: W: no-manual-page-for-binary band_plot.x quantum-espresso.i686: W: no-manual-page-for-binary rrkj2upf.x quantum-espresso.i686: W: no-manual-page-for-binary bands_FS.x quantum-espresso.i686: W: no-manual-page-for-binary cp.x quantum-espresso.i686: W: no-manual-page-for-binary kvecs_FS.x quantum-espresso.i686: W: no-manual-page-for-binary ld1.x quantum-espresso.i686: W: no-manual-page-for-binary interpolate.x quantum-espresso.i686: W: no-manual-page-for-binary ncpp2upf.x quantum-espresso.i686: W: no-manual-page-for-binary fd.x quantum-espresso.i686: W: no-manual-page-for-binary lambda.x quantum-espresso.i686: W: no-manual-page-for-binary read_upf_tofile.x quantum-espresso.i686: W: no-manual-page-for-binary bgw2pw.x quantum-espresso.i686: W: no-manual-page-for-binary dos.x quantum-espresso.i686: W: no-manual-page-for-binary uspp2upf.x quantum-espresso.i686: W: no-manual-page-for-binary oldcp2upf.x quantum-espresso.i686: W: no-manual-page-for-binary iotk.x quantum-espresso.i686: W: no-manual-page-for-binary kpoints.x quantum-espresso.i686: W: no-manual-page-for-binary gww_fit.x quantum-espresso.i686: W: no-manual-page-for-binary pw.x quantum-espresso.i686: W: no-manual-page-for-binary cpmd2upf.x quantum-espresso.i686: W: no-manual-page-for-binary iotk_print_kinds.x quantum-espresso.i686: W: no-manual-page-for-binary dist.x quantum-espresso.i686: W: no-manual-page-for-binary pwi2xsf.x quantum-espresso.i686: W: no-manual-page-for-binary fpmd2upf.x quantum-espresso.i686: W: no-manual-page-for-binary q2trans.x quantum-espresso.i686: W: no-manual-page-for-binary fhi2upf.x quantum-espresso.i686: W: no-manual-page-for-binary plotband.x quantum-espresso.i686: W: no-manual-page-for-binary dynmat.x quantum-espresso.i686: W: no-manual-page-for-binary wannier_plot.x quantum-espresso.i686: W: no-manual-page-for-binary vdb2upf.x quantum-espresso.i686: W: no-manual-page-for-binary d3.x quantum-espresso.i686: W: no-manual-page-for-binary ph.x quantum-espresso.i686: W: no-manual-page-for-binary head.x quantum-espresso.i686: W: no-manual-page-for-binary q2trans_fd.x quantum-espresso.i686: W: no-manual-page-for-binary fqha.x quantum-espresso.i686: W: no-manual-page-for-binary pw_export.x quantum-espresso.i686: W: no-manual-page-for-binary pmw.x quantum-espresso.i686: W: no-manual-page-for-binary iotk quantum-espresso.i686: W: no-manual-page-for-binary matdyn.x quantum-espresso.i686: W: no-manual-page-for-binary turbo_spectrum.x quantum-espresso.i686: W: no-manual-page-for-binary manypw.x quantum-espresso.i686: W: no-manual-page-for-binary turbo_davidson.x quantum-espresso.i686: W: no-manual-page-for-binary q2qstar.x quantum-espresso.i686: W: no-manual-page-for-binary wfdd.x quantum-espresso.i686: W: no-manual-page-for-binary pw2gw.x quantum-espresso.i686: W: no-manual-page-for-binary generate_vdW_kernel_table.x quantum-espresso.i686: W: no-manual-page-for-binary initial_state.x quantum-espresso.i686: W: no-manual-page-for-binary gww.x quantum-espresso.i686: W: no-manual-page-for-binary average.x quantum-espresso.i686: W: no-manual-page-for-binary phcg.x quantum-espresso.i686: W: no-manual-page-for-binary neb.x quantum-espresso.i686: W: no-manual-page-for-binary plotproj.x quantum-espresso.i686: W: no-manual-page-for-binary pw2bgw.x quantum-espresso.i686: W: no-manual-page-for-binary virtual.x quantum-espresso.i686: W: no-manual-page-for-binary turbo_lanczos.x quantum-espresso.i686: W: no-manual-page-for-binary epsilon.x quantum-espresso.i686: W: no-manual-page-for-binary casino2upf.x quantum-espresso.i686: W: no-manual-page-for-binary projwfc.x quantum-espresso.i686: W: no-manual-page-for-binary pwcond.x quantum-espresso.i686: W: no-manual-page-for-binary sumpdos.x quantum-espresso.i686: W: no-manual-page-for-binary upf2casino.x quantum-espresso.i686: W: no-manual-page-for-binary importexport_binary.x quantum-espresso.i686: W: no-manual-page-for-binary q2r.x quantum-espresso.i686: W: no-manual-page-for-binary turbo_eels.x quantum-espresso.i686: W: no-manual-page-for-binary spectra_manipulation.x quantum-espresso.i686: W: no-manual-page-for-binary wfck2r.x quantum-espresso.i686: W: no-manual-page-for-binary bands.x quantum-espresso.i686: W: no-manual-page-for-binary plan_avg.x quantum-espresso.i686: W: no-manual-page-for-binary xspectra.x quantum-espresso.i686: W: no-manual-page-for-binary ev.x quantum-espresso.src: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso.src: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso.src: W: spelling-error %description -l en_US iotk -> iota quantum-espresso.src: W: spelling-error %description -l en_US executables -> executable, executable s, executrices quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dot.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dav_main.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_calc_dens_eels_nc.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_ortho.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dealloc.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/PW/tools/dist.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/PW/src/h_epsi_her_set.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_setup_nscf.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_lanczos.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_solve_e.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_calc_dens.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_variables.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_calc_dens_eels.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/stop_lr.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_restart.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_sm1_psi.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/sd0psi.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/bcast_lr_input.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/Modules/xc_rVV10.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_read_wf_eels_periodic.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_apply_liouvillian_eels.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/PW/src/generate_rVV10_kernel_table.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_sym_eels.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_smallgq.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_test_restart.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_psym_eels.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_summary.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dvpsi_eels.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dvpsi_e.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_main.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_addusddens.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/XSpectra/src/banner_xspectra.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_dv_setup.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_read_wf.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_run_nscf.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_readin.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/print_clock_lr.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_read_d0psi.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_write_restart.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_alloc_init.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_apply_liouvillian.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/PW/src/rdiagh.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/tools/tddfpt_calculate_spectrum.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_set_boxes_density.f90 quantum-espresso-debuginfo.i686: E: script-without-shebang /usr/src/debug/espresso-5.2.1/TDDFPT/src/lr_eels_main.f90 quantum-espresso-devel.i686: W: summary-not-capitalized C quantum-espresso - devel quantum-espresso-devel.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-devel.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-devel.i686: W: only-non-binary-in-usr-lib quantum-espresso-devel.i686: W: no-documentation quantum-espresso-devel.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-devel/License quantum-espresso-mpich.i686: W: summary-not-capitalized C quantum-espresso - mpich version quantum-espresso-mpich.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-mpich.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-mpich.i686: W: no-documentation quantum-espresso-mpich.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-mpich/License quantum-espresso-mpich-devel.i686: W: summary-not-capitalized C quantum-espresso - devel mpich version quantum-espresso-mpich-devel.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-mpich-devel.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-mpich-devel.i686: W: only-non-binary-in-usr-lib quantum-espresso-mpich-devel.i686: W: no-documentation quantum-espresso-mpich-devel.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-mpich-devel/License quantum-espresso-mpich-static.i686: W: summary-not-capitalized C quantum-espresso - static libraries mpich version quantum-espresso-mpich-static.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-mpich-static.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-mpich-static.i686: W: no-documentation quantum-espresso-mpich-static.i686: W: spurious-executable-perm /usr/lib/mpich/lib/libiotk.a quantum-espresso-mpich-static.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-mpich-static/License quantum-espresso-openmpi.i686: W: summary-not-capitalized C quantum-espresso - openmpi version quantum-espresso-openmpi.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-openmpi.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-openmpi.i686: W: no-documentation quantum-espresso-openmpi.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-openmpi/License quantum-espresso-openmpi-devel.i686: W: summary-not-capitalized C quantum-espresso - devel openmpi version quantum-espresso-openmpi-devel.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-openmpi-devel.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-openmpi-devel.i686: W: only-non-binary-in-usr-lib quantum-espresso-openmpi-devel.i686: W: no-documentation quantum-espresso-openmpi-devel.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-openmpi-devel/License quantum-espresso-openmpi-static.i686: W: summary-not-capitalized C quantum-espresso - static libraries openmpi version quantum-espresso-openmpi-static.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-openmpi-static.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-openmpi-static.i686: W: no-documentation quantum-espresso-openmpi-static.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-openmpi-static/License quantum-espresso-openmpi-static.i686: W: spurious-executable-perm /usr/lib/openmpi/lib/libiotk.a quantum-espresso-static.i686: W: summary-not-capitalized C quantum-espresso - static libraries quantum-espresso-static.i686: W: spelling-error %description -l en_US nanoscale -> nanosecond quantum-espresso-static.i686: W: spelling-error %description -l en_US pseudopotentials -> pseudo potentials, pseudo-potentials, potentials quantum-espresso-static.i686: W: no-documentation quantum-espresso-static.i686: E: incorrect-fsf-address /usr/share/licenses/quantum-espresso-static/License quantum-espresso-static.i686: W: spurious-executable-perm /usr/lib/libiotk.a 11 packages and 0 specfiles checked; 54 errors, 128 warnings. rpmlint /var/lib/mock/fedora-rawhide-i386/result/*rpm 75.71s user 5.11s system 86% cpu 1:33.30 total Mostly false positives, nothing important. Package is APPROVED.
Thanks. For iotk, let's see if espresso developers react to the presence of the RPM, I drop them a message. Maybe they consider splitting iotk. I have verified that parallel runs are OK with openmpi and mpich on CentOS 7. I mean the output from an expresso run shows the expected number of N processes and there is no N-tupling of lines in the output, which would be the case if one runs a serial code with mpiexec.
Package request has been approved: https://admin.fedoraproject.org/pkgdb/package/quantum-espresso
FYI: I'm disabling armv7hl build because it took almost one week on Fedora 24 without finishing http://koji.fedoraproject.org/koji/taskinfo?taskID=12573023 Moreover it looks like the build has been restarted 4 times.
Actually this is the test-suite that takes so long on armv7hl. The test-suite finished in serial (1-2 days), but never finished with openmpi and did not even reach mpich. For comparison the whole build+test-suite on x86_64 %{ix86} took 2 days https://koji.fedoraproject.org/koji/taskinfo?taskID=12593524 (still several times longer compared to a build on my laptop).
quantum-espresso-5.3.0-5.el6 has been submitted as an update to Fedora EPEL 6. https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-ddb5c45a81
quantum-espresso-5.3.0-5.fc22 has been submitted as an update to Fedora 22. https://bodhi.fedoraproject.org/updates/FEDORA-2016-65540a772a
quantum-espresso-5.3.0-5.fc23 has been submitted as an update to Fedora 23. https://bodhi.fedoraproject.org/updates/FEDORA-2016-31f2840d93
quantum-espresso-5.3.0-5.el6 has been pushed to the Fedora EPEL 6 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-ddb5c45a81
quantum-espresso-5.3.0-5.fc23 has been pushed to the Fedora 23 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2016-31f2840d93
quantum-espresso-5.3.0-5.fc22 has been pushed to the Fedora 22 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2016-65540a772a
Message sent to quantum-espresso developers: http://qe-forge.org/pipermail/pw_forum/2016-January/108949.html
quantum-espresso-5.3.0-7.el7 has been submitted as an update to Fedora EPEL 7. https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-7d9b8990e3
quantum-espresso-5.3.0-5.fc23 has been pushed to the Fedora 23 stable repository. If problems still persist, please make note of it in this bug report.
quantum-espresso-5.3.0-5.fc22 has been pushed to the Fedora 22 stable repository. If problems still persist, please make note of it in this bug report.
quantum-espresso-5.3.0-7.el7 has been pushed to the Fedora EPEL 7 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-7d9b8990e3
quantum-espresso-5.3.0-5.el6 has been pushed to the Fedora EPEL 6 stable repository. If problems still persist, please make note of it in this bug report.
quantum-espresso-5.3.0-8.el7 has been submitted as an update to Fedora EPEL 7. https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-72ad34fd5f
quantum-espresso-5.3.0-8.el7 has been pushed to the Fedora EPEL 7 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2016-72ad34fd5f
quantum-espresso-5.3.0-8.el7 has been pushed to the Fedora EPEL 7 stable repository. If problems still persist, please make note of it in this bug report.