[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 Neal Becker ndbeck...@gmail.com changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 Thorsten Leemhuis fed...@leemhuis.info changed: What|Removed |Added Blocks|33 | --- Comment #33 from Thorsten Leemhuis fed...@leemhuis.info 2009-01-27 19:19:12 --- CVS done -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #32 from Neal Becker ndbeck...@gmail.com 2009-01-17 00:30:50 --- Updated sources: http://nbecker.dyndns.org:8080/RPM/blcr.spec http://nbecker.dyndns.org:8080/RPM/blcr-0.8.0-1.fc10.src.rpm -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #29 from Kevin Kofler kevin.kof...@chello.at 2009-01-06 13:40:08 --- Just use ExclusiveArch: i586 ppc ppc64 arm. Or use i486 or i686 instead of i586, depending on how old stuff you want to support, but the point is that one is enough. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #31 from Dominik 'Rathann' Mierzejewski r...@greysector.net 2009-01-06 23:51:51 --- (In reply to comment #30) Not directly related to this bug, but can someone give a brief explanation what i486, i586 are for? Is i686 good for everything or is it good for Pentium X and above only? If so, X=? and what is the athlon correspondence? i686 is, AFAIK, anything that glibc considers i686 (i.e. uname -m == i686). I'd say that's Pentium Pro and above (i.e. Pentium II/III/4). From AMD camp that's Athlon and above and from VIA it's VIA C3 and above (even though C3 doesn't have the cmov instruction, which means you shouldn't use -march=i686 when compiling stuff for it; C3-2 does have it though). HTH -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 Neal Becker ndbeck...@gmail.com changed: What|Removed |Added Blocks||33 --- Comment #26 from Neal Becker ndbeck...@gmail.com 2008-12-23 18:03:11 --- Package CVS request == Package Name: blcr Short Description: Berkeley Lab Checkpoint/Restart for Linux (BLCR) This package implements system-level checkpointing of scientific applications in a manner suitable for implementing preemption, migration and fault recovery by a batch scheduler. Owners: ndbeck...@gmail.com Branches: F9, F10 InitialCC: -- License tag: free -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #18 from Kevin Kofler kevin.kof...@chello.at 2008-12-22 10:43:47 --- Can't the bug which requires -fno-stack-protector be fixed?! IMHO this is a blocker, packages should NEVER use -fno-stack-protector, it's an invitation for crackers to exploit any buffer overflows, and the fact that it's needed in the first place strongly points to such buffer overflows being present. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #19 from Kevin Kofler kevin.kof...@chello.at 2008-12-22 10:45:09 --- Or let's say it differently: WHY is -fno-stack-protector needed? Where's the evidence that the stack protector isn't catching a legitimate buffer overflow? -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #20 from Neal Becker ndbeck...@gmail.com 2008-12-23 00:47:02 --- OK, I believe we have all issues addressed. * Static libs in static package * Patched to eliminate -fno-stack-protector Builds, links, passes test suite on x86_64 (which is the only platform I have to test it on). Note: I did _not_ enable check to run the testsuite, because you have to load the kernel module to run the testsuite. http://nbecker.dyndns.org:8080/RPM/blcr.spec http://nbecker.dyndns.org:8080/RPM/blcr-kmod.spec http://nbecker.dyndns.org:8080/RPM/blcr-0.8.0_b5-4.fc10.src.rpm http://nbecker.dyndns.org:8080/RPM/blcr-kmod-0.8.0_b5-4.fc10.src.rpm -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #21 from Orcan Ogetbil oget.fed...@gmail.com 2008-12-23 01:02:47
---
* Package doesn't build with mock. It fails with:
Patch #1 (blcr-stackcheck.patch):
+ /bin/cat /builddir/build/SOURCES/blcr-stackcheck.patch
+ /usr/bin/patch -s -p0 --fuzz=0
+ autoreconf --force --install
Can't exec aclocal: Permission denied at /usr/share/autoconf/Autom4te
FileUtils.pm line 326.
autoreconf: failed to run aclocal: Permission denied
* Still, the file README.devel is not and should be packaged (in the devel
subpackage).
* The %if %{build_static} phrase inside %files static must go up, right
before %package static. Otherwise when %{build_static} is disabled you'll
get an empty static subpackage.
* The commented out parts of the SPEC file need to be removed if there's no
legitimate reason to keep them. If there's a legitimate reason that should be
explained within the SPEC file.
* No macros with single % in a commented-out part is allowed. Use %%
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #23 from Neal Becker ndbeck...@gmail.com 2008-12-23 02:46:02 ---
(In reply to comment #21)
* Package doesn't build with mock. It fails with:
Patch #1 (blcr-stackcheck.patch):
+ /bin/cat /builddir/build/SOURCES/blcr-stackcheck.patch
+ /usr/bin/patch -s -p0 --fuzz=0
+ autoreconf --force --install
Can't exec aclocal: Permission denied at /usr/share/autoconf/Autom4te
FileUtils.pm line 326.
autoreconf: failed to run aclocal: Permission denied
OK, fixed. The patch is requires running reconfigure, so BR automake, libtools
* Still, the file README.devel is not and should be packaged (in the devel
subpackage).
I think that's gone, isn't it?
* The %if %{build_static} phrase inside %files static must go up, right
before %package static. Otherwise when %{build_static} is disabled you'll
get an empty static subpackage.
Done.
* The commented out parts of the SPEC file need to be removed if there's no
legitimate reason to keep them. If there's a legitimate reason that should be
explained within the SPEC file.
Done.
* No macros with single % in a commented-out part is allowed. Use %%
Done.
http://nbecker.dyndns.org:8080/RPM/blcr.spec
http://nbecker.dyndns.org:8080/RPM/blcr-kmod.spec
http://nbecker.dyndns.org:8080/RPM/blcr-0.8.0-0.1.b5.fc10.src.rpm
http://nbecker.dyndns.org:8080/RPM/blcr-kmod-0.8.0-0.1.b5.fc10.src.rpm
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 Orcan Ogetbil oget.fed...@gmail.com changed: What|Removed |Added Blocks|2 |4 Status|NEW |ASSIGNED --- Comment #24 from Orcan Ogetbil oget.fed...@gmail.com 2008-12-23 03:35:23 --- (In reply to comment #23) (In reply to comment #21) * Still, the file README.devel is not and should be packaged (in the devel subpackage). I think that's gone, isn't it? * Yes. Actually, that's the problem. Read again, I'm saying, - the README.devel file is not packaged - the README.devel file should be packaged. Sorry if I brought confusion. * Other than these, it would be nice if you explain in the SPEC file what each patch does (e.g. Patch1) Please do these little changes before you commit. I think the package is good to go now. -- This package (blcr) is APPROVED by oget for rpmfusion-free -- For the kmod package, I believe Thorsten will do the review. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #16 from Neal Becker ndbeck...@gmail.com 2008-12-22 00:03:21 --- I believe all issues are addressed. According to upstream, building of some sources with -fno-stack-protector is required on some platform. Please see: http://nbecker.dyndns.org:8080/RPM/blcr.spec http://nbecker.dyndns.org:8080/RPM/blcr-0.8.0_b5-2.fc10.src.rpm -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #17 from Orcan Ogetbil oget.fed...@gmail.com 2008-12-22 07:47:28 --- Almost there. You missed this part: * The file README.devel is not and should be packaged. My understanding is, you don't want to completely remove the static library bits from SPEC. If they are to be included (either disabled or enabled) they need to obey the guidelines, i.e. they need to go to a -static subpackage. Please read: http://fedoraproject.org/wiki/Packaging/Guidelines#Exclusion_of_Static_Libraries The way it is right now bothers me. It is disabled in the SPEC, but enabling them will put the static libraries into the devel package, which is wrong. Maybe you should extend the if-build_static-clause and define the -static subpackage inside it. Then if someone wants to build the static bits enabled, the static libraries will get into the correct subpackage. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #13 from Neal Becker ndbeck...@gmail.com 2008-12-18 16:45:45 ---
(In reply to comment #10)
This package surely needs some work. To start with:
* mock build fails on my x86_64. This is because you are trying to build and
include 32 bit libraries in a 64 bit package, which is not allowed. If one
needs 32 bit libraries (s)he can install blcr-libs.i386 in addition to
blcr_libs.x86_64 . So you should remove the libdir32 bits from the SPEC
file.
Fixed
* Leave a comment in the SPEC file for why you are using ExclusiveArch.
Done
* Try to avoid mixed ${ } %{_ } notation
Do you mean:
chrpath -d ${RPM_BUILD_ROOT}/%{_bindir}/cr_checkpoint
Suggested alternative?
* BR: perl and sed are not required since they are in the minimum build
environment.
Done
* Please remove the static library bits from the SPEC file.
I assume you mean to unconditionally build static libs for devel.
Done.
* rpmlint complains:
blcr-devel.x86_64: W: no-documentation
blcr-testsuite.x86_64: W: no-documentation
blcr-testsuite.x86_64: E: script-without-shebang
/usr/libexec/blcr-testsuite/shellinit
For the first two, at least put the license file(s) in those packages.
The last one is actually about an empty files. Well it is not empty but when
you open it, it says #empty. Do you think we should include that file?
Fixed
* Patches should be explained and be submitted to upstream if they are not
strictly Fedora specific.
Done.
* The file tests/CountingApp.class is binary and should be removed during
%prep
It isn't installed, do we care?
* The file README.devel is not and should be packaged.
Why?
* Buildroot should be one of these:
%(mktemp -ud %{_tmppath}/%{name}-%{version}-%{release}-XX)
%{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
%{_tmppath}/%{name}-%{version}-%{release}-root
Done.
* Why do you have:
# Ensure we don't build for a i386
%ifarch i386
set +x
echo
==
echo ERROR: Cannot build BLCR for a generic i386. 2
echo ERROR: Add \--target `uname -p`\ (or similar) to the rpmbuild
command line. 2
echo
==
exit 1
%endif
in the SPEC file? Just remove i386 from ExclusiveArch and you should be fine.
Fixed.
* Please use
%post libs -p /sbin/ldconfig
%postun libs -p /sbin/ldconfig
Afaik, they'll work more efficient.
Done.
* We prefer %defattr(-,root,root,-)
Done.
* Each package must consistently use macros, as described in the macros
section
of Fedora Packaging Guidelines . Avoid inconsistencies such as:
%clean
rm -rf ${RPM_BUILD_ROOT}
%install
rm -rf $RPM_BUILD_ROOT
Done.
* Disttag is missing.
What is this?
* The Fedora-specific compilation flag -fstack-protector is not passed to the
compiler. For a list of flags that should be passed to the compiler, please do
a
rpm --eval %optflags
I believe all flags are passed, because %configure is used. I just tested it,
and I believe -fstack-protector is passed.
* Parallel make must be supported whenever possible. If it is not supported,
this should be noted in the SPEC file as a comment.
Seems to break on this package, comment added.
* Shall we package the examples, tests directories?
I think it's good to have the testsuite.
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #14 from Orcan Ogetbil orcanba...@yahoo.com 2008-12-18 17:40:12
---
(In reply to comment #13)
(In reply to comment #10)
* Try to avoid mixed ${ } %{_ } notation
Do you mean:
chrpath -d ${RPM_BUILD_ROOT}/%{_bindir}/cr_checkpoint
Suggested alternative?
$RPM_BUILD_ROOT or possibly just %{buildroot} . Actually, on a second thought,
this particular one is no big deal, you can leave it as is. Just try to stay
as consistent as possible with the macros in the future.
* Please remove the static library bits from the SPEC file.
I assume you mean to unconditionally build static libs for devel.
Done.
No, I meant: no static libraries at all.
Well if there's a definite need for static libraries. They should be built and
put in a -static subpackage.
* The file tests/CountingApp.class is binary and should be removed during
%prep
It isn't installed, do we care?
Yes, we do. It is a common practice to remove the precompiled binaries in %prep
* The file README.devel is not and should be packaged.
Why?
It contains information for people who might want to develop this software.
* Disttag is missing.
What is this?
Disttag is: %{?dist}
You usually add it to the end of Release tag.
For example:
Version: 2.5
Release: 3{?dist}
* The Fedora-specific compilation flag -fstack-protector is not passed to
the
compiler. For a list of flags that should be passed to the compiler, please
do
a
rpm --eval %optflags
Search for -fno-stack-protector in the build log. During the compilation of
certain files, that one is passed instead of -fstack-protector .
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #15 from Kevin Kofler kevin.kof...@chello.at 2008-12-18 23:57:27 --- The disttag is not mandatory. On the other hand, not using a disttag means you have to make sure the Release tag is distinct on each branch by hand, so it's strongly recommended to use one. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #11 from Neal Becker ndbeck...@gmail.com 2008-12-17 13:25:29 ---
(In reply to comment #10)
This package surely needs some work. To start with:
* mock build fails on my x86_64. This is because you are trying to build and
include 32 bit libraries in a 64 bit package, which is not allowed. If one
needs 32 bit libraries (s)he can install blcr-libs.i386 in addition to
blcr_libs.x86_64 . So you should remove the libdir32 bits from the SPEC
file.
* Leave a comment in the SPEC file for why you are using ExclusiveArch.
* Try to avoid mixed ${ } %{_ } notation
* BR: perl and sed are not required since they are in the minimum build
environment.
* Please remove the static library bits from the SPEC file.
* rpmlint complains:
blcr-devel.x86_64: W: no-documentation
blcr-testsuite.x86_64: W: no-documentation
blcr-testsuite.x86_64: E: script-without-shebang
/usr/libexec/blcr-testsuite/shellinit
For the first two, at least put the license file(s) in those packages.
The last one is actually about an empty files. Well it is not empty but when
you open it, it says #empty. Do you think we should include that file?
* Patches should be explained and be submitted to upstream if they are not
strictly Fedora specific.
* The file tests/CountingApp.class is binary and should be removed during
%prep
* The file README.devel is not and should be packaged.
* Buildroot should be one of these:
%(mktemp -ud %{_tmppath}/%{name}-%{version}-%{release}-XX)
%{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
%{_tmppath}/%{name}-%{version}-%{release}-root
* Why do you have:
# Ensure we don't build for a i386
%ifarch i386
set +x
echo
==
echo ERROR: Cannot build BLCR for a generic i386. 2
echo ERROR: Add \--target `uname -p`\ (or similar) to the rpmbuild
command line. 2
echo
==
exit 1
%endif
in the SPEC file? Just remove i386 from ExclusiveArch and you should be fine.
* Please use
%post libs -p /sbin/ldconfig
%postun libs -p /sbin/ldconfig
Afaik, they'll work more efficient.
* We prefer %defattr(-,root,root,-)
* Each package must consistently use macros, as described in the macros
section
of Fedora Packaging Guidelines . Avoid inconsistencies such as:
%clean
rm -rf ${RPM_BUILD_ROOT}
%install
rm -rf $RPM_BUILD_ROOT
* Disttag is missing.
* The Fedora-specific compilation flag -fstack-protector is not passed to the
compiler. For a list of flags that should be passed to the compiler, please do
a
rpm --eval %optflags
* Parallel make must be supported whenever possible. If it is not supported,
this should be noted in the SPEC file as a comment.
* Shall we package the examples, tests directories?
Thank you. I am working with upstream on these.
The 32bit is the most challenge. I think what we want is that we wind up with
seperate 32bit and 64bit libs packages, blcr-libs.x86_64 and blcr-libs.i386.
Consistent with other multi-arch packages, we want 32bit libs available on
64bit arch, but not installed by default.
What is the standard way to setup srpm to produce this result?
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #12 from Thorsten Leemhuis fed...@leemhuis.info 2008-12-17 15:04:41 --- (In reply to comment #11) What is the standard way to setup srpm to produce this result? That will be done automatically by the push scripts if you have a -devel and a -libs subpackage -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #8 from Orcan Ogetbil orcanba...@yahoo.com 2008-12-15 18:03:07 --- The links seem to be broken. Did you just do an update on your server? -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #9 from Neal Becker ndbeck...@gmail.com 2008-12-15 18:53:43 --- (In reply to comment #8) The links seem to be broken. Did you just do an update on your server? Yes, sorry. Moved to a new server. Please try again. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
--- Comment #10 from Orcan Ogetbil orcanba...@yahoo.com 2008-12-16 06:48:50
---
This package surely needs some work. To start with:
* mock build fails on my x86_64. This is because you are trying to build and
include 32 bit libraries in a 64 bit package, which is not allowed. If one
needs 32 bit libraries (s)he can install blcr-libs.i386 in addition to
blcr_libs.x86_64 . So you should remove the libdir32 bits from the SPEC file.
* Leave a comment in the SPEC file for why you are using ExclusiveArch.
* Try to avoid mixed ${ } %{_ } notation
* BR: perl and sed are not required since they are in the minimum build
environment.
* Please remove the static library bits from the SPEC file.
* rpmlint complains:
blcr-devel.x86_64: W: no-documentation
blcr-testsuite.x86_64: W: no-documentation
blcr-testsuite.x86_64: E: script-without-shebang
/usr/libexec/blcr-testsuite/shellinit
For the first two, at least put the license file(s) in those packages.
The last one is actually about an empty files. Well it is not empty but when
you open it, it says #empty. Do you think we should include that file?
* Patches should be explained and be submitted to upstream if they are not
strictly Fedora specific.
* The file tests/CountingApp.class is binary and should be removed during %prep
* The file README.devel is not and should be packaged.
* Buildroot should be one of these:
%(mktemp -ud %{_tmppath}/%{name}-%{version}-%{release}-XX)
%{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
%{_tmppath}/%{name}-%{version}-%{release}-root
* Why do you have:
# Ensure we don't build for a i386
%ifarch i386
set +x
echo
==
echo ERROR: Cannot build BLCR for a generic i386. 2
echo ERROR: Add \--target `uname -p`\ (or similar) to the rpmbuild
command line. 2
echo
==
exit 1
%endif
in the SPEC file? Just remove i386 from ExclusiveArch and you should be fine.
* Please use
%post libs -p /sbin/ldconfig
%postun libs -p /sbin/ldconfig
Afaik, they'll work more efficient.
* We prefer %defattr(-,root,root,-)
* Each package must consistently use macros, as described in the macros section
of Fedora Packaging Guidelines . Avoid inconsistencies such as:
%clean
rm -rf ${RPM_BUILD_ROOT}
%install
rm -rf $RPM_BUILD_ROOT
* Disttag is missing.
* The Fedora-specific compilation flag -fstack-protector is not passed to the
compiler. For a list of flags that should be passed to the compiler, please do
a
rpm --eval %optflags
* Parallel make must be supported whenever possible. If it is not supported,
this should be noted in the SPEC file as a comment.
* Shall we package the examples, tests directories?
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 Orcan Ogetbil [EMAIL PROTECTED] changed: What|Removed |Added CC||[EMAIL PROTECTED] --- Comment #6 from Orcan Ogetbil [EMAIL PROTECTED] 2008-11-30 05:05:19 --- I'll review this package but only if you address those cosmetic issues, which are required by the Fedora guidelines and here at rpmfusion, we stay consistent with them. Secondly, how is this package supposed to be built? Are the 32bit libraries supposed to be compiled in 64bit systems? Remember that the SPEC file has to be ready to go via 'rpmbuild -ba blcr.spec' in all systems. -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19 --- Comment #5 from Neal Becker [EMAIL PROTECTED] 2008-03-01 15:32:17 --- Updated from upsteam. Add BR chrpath Kill .la files (we don't install them, correct?) I did not address the cosmetic issues that were raised in the previous comment. Some of the complexity is because the upstream supplied an srpm, which I simplified already. I'd rather not deviate too much without good reason. http://nbecker.dyndns.org:8080/RPM/blcr-0.6.5-2.src.rpm http://nbecker.dyndns.org:8080/RPM/blcr-kmod-0.6.5-1.fc8.src.rpm http://nbecker.dyndns.org:8080/RPM/blcr-kmod.spec http://nbecker.dyndns.org:8080/RPM/blcr.spec -- Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the assignee for the bug.
[Bug 19] Review request: blcr - Berkeley Lab Checkpoint/Restart for Linux
http://bugzilla.rpmfusion.org/show_bug.cgi?id=19
David Timms [EMAIL PROTECTED] changed:
What|Removed |Added
CC||[EMAIL PROTECTED]
--- Comment #4 from David Timms [EMAIL PROTECTED] 2008-02-29 15:41:58 ---
Neil: cursory look at https://nbecker.dyndns.org/RPM/blcr.spec
1.
%install
rm -rf $RPM_BUILD_ROOT
noticed missing / inconsistent braces.
2. chrpath -d ${RPM_BUILD_ROOT}/%{_bindir}/cr_checkpoint
mixed ${X} and %{_Y} should be changed to use one consistent method throughout:
3. No static library builds - without those options the spec would become a lot
simpler. {does that apply to this type of package ?}
4. ExclusiveArch: i386 i486 i586 i686 athlon x86_64 ppc64
Leave comment for using exclude arch, ie why ?
I am yet to try building it.
--
Configure bugmail: http://bugzilla.rpmfusion.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are the assignee for the bug.
