Quite recently a few macros were added to fedora-release: %dist_vendor,
%dist_name, %dist_home_url and %dist_bug_report_url. These will
eventually be documented in the Fedora packaging guidelines and you can
see the initial PR at
https://pagure.io/packaging-committee/pull-request/1198
I was
> Troy Dawson writes:
> If I remember right, the spec file name needs to be in the format
> .spec Thus, the spec file needs to be
> openssl3.spec, and thus you aren't really renaming it. :)
Yes, assuming that EPEL doesn't deviate:
> "SJS" == Stephen John Smoogen writes:
SJS> Can koschei be limited to a single architecture or does it need to
SJS> build against all targets? I am worried about the number of s390
SJS> builds we may be adding.
Currently it only does builds for aarch64, ppc64le and x86_64, so it
does at
> "PW" == Phil Wyett writes:
PW> The '.1' should go before the %{?dist} tag in this case so to not
PW> confuse as the outputted 'el8.1' does.
This is not true:
The issues you show don't appear to have anything to do with EPEL. The
certbot package simply requires /usr/sbin/restorecon and
/usr/sbin/semanage; yum isn't able to install those because it appears
that there is some problem with the base (Centos) repositories. Please
make sure that you are
> "DD" == David Dykstra writes:
DD> Does this also imply a new fuse3 package in pagure & bodhi? Or
DD> could it still be part of the fuse package but have a separate
DD> fuse3.spec?
It would have to be a separate package repository and receive separate
updates. It is not possible for a
The denyhosts software has only been lightly maintained in EPEL for many
years; the EPEL6 branch could have been considered to be unmaintained.
It does not properly support modern system features such as systemd or
firewalld and generally expects to be able to make use of the hosts.deny
file. It
> "NG" == Neal Gompa writes:
NG> Wait, we can do that? I thought we couldn't use the exception
NG> process for this?
Well, the idea is that you don't need a separate review just to import a
different version of the same package. So foo and foo1.2 (the 1.2
version) or python-abc and
> "OP" == Orion Poplawski writes:
OP> - Can we make epel7-py36 branches, and somehow have
OP> %python3_version, et. al. be 3.6 for those builds?
I can't think of any way to do that without extra magic. And if you
require something in the spec, you might as well just hardcode it.
OP> - Can
> "RI" == Roman Iuvshyn writes:
RI> I have a question related to *python2-jenkins-job-builder
RI> *package, who maintains this?
https://src.fedoraproject.org/rpms/python-jenkins-job-builder shows all
of the maintainers.
RI> currently available version in epel is pretty old and I wonder if
> "SJS" == Stephen John Smoogen writes:
SJS> This looks to be one of those packages which are only in EPEL for
SJS> users of the aarch64 and ppc64 architectures.
I wonder how you can tell. The specfile doesn't indicate anything about
it.
- J<
I asked on IRC and mboddu had a look; somehow the retirement didn't
automatically filter into the PDC (which keeps all sorts of metadata
about packages) and so nothing else in the system noticed. There were
sporadic problems with this in the past (which have been fixed for a
while) but a few
> "NdV" == Niels de Vos writes:
NdV> I think I followed [1] correctly, but the package is still
NdV> there. Could someone have a look at the commit [2] with
NdV> dead.package in it and advise me what I did wrong, or what steps I
NdV> have missed?
You linked to the document, but did you read
Also, when I look at the four updates you listed, I see that they are
all recent and that they are all currently in stable, not testing.
Perhaps you are pulling from a mirror which is out of date and still has
old data for epel-testing?
If you do 'yum updateinfo list', I believe it will tell you
> "p" == postmaster writes:
[...]
p> consider re-running the same command with --verbose to see the exact
p> data that caused the conflict.
Could perhaps you re-run the same command with --verbose so that we can
see the exact data that caused the conflict?
I run yum-cron with epel-testing
Just a note that I've backported the %set_build_flags macro from Fedora
to EPEL6 and EPEL7. Updates are at:
* https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2018-99aa68bf61 (EPEL7)
* https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2018-6b0faf2b25 (EPEL6)
Buildroot overrides have been
The package should appear in the repositories with the next EPEL7
compose.
- J<
___
epel-devel mailing list -- epel-devel@lists.fedoraproject.org
To unsubscribe send an email to epel-devel-le...@lists.fedoraproject.org
> "GM" == Germano Massullo writes:
GM> but esteidcerts package is in EPEL7 *stable* repositories [2], so
GM> why do I get such message?
That update report indicates that it was pushed to stable, but as far as
I can tell it does not actually appear in the stable
The initial set of stub python2-* packages I created have now made their
way to EPEL6 and EPEL7 stable, so packages can now depend on
python2-setuptools, python2-six, python2-pytest and python2-sphinx
in all releases without having to add conditionals for EPEL.
Feel free to suggest additional
All of the initial run of packages have been submitted now. See
https://fedoraproject.org/wiki/User:Tibbs/EPELPythonStubPackages for the
complete list of builds.
Please test and give karma. Thanks,
- J<
___
epel-devel mailing list --
It would of course be better if I actually provided links to the
updates:
python2-setuptools (EPEL6):
https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2018-92c3e1b0e4
python2-sphinx (EPEL7):
https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2018-9cd64dfc3c
python2-pytest (EPEL7):
(This is mostly a duplicate of a post I sent to devel@. I wanted to
alert epel-devel@ but didn't want to crosspost.)
Following my proposal in
https://fedoraproject.org/wiki/User:Tibbs/EPELPythonStubPackages which
met with favor from a number of folks here, I went ahead and set up four
dummy
> "SJS" == Stephen John Smoogen writes:
SJS> OK how can we better explain this in the future?
I really tried, in the "Can I rely on these packages?" section of the
EPEL wiki page:
https://fedoraproject.org/wiki/EPEL#Can_I_rely_on_these_packages.3F
Someone already quoted
> "KF" == Kevin Fenzi writes:
KF> I don't think we have any way to find out the version of a package
KF> in all channels. At least I don't know of such a way. So, I think we
KF> should concern ourselves with only the channels that EPEL says it
KF> will try and not conflict
One of things I've invested some time towards over the years is cutting
down on the amount of %if spaghetti needed to share spec files between
Fedora and EPEL. Earlier in the discussion about the mass python-X ->
python2-X move I volunteered to maintain a number of "stub packages" in
EPEL which
I've been doing a lot of packaging work on the Fedora version of the
cyrus-imapd package, including a lot of work with upstream to do things
like run their external test suite as part of the package build process,
and fix issues that only show up on 32-bit or big-endian architectures.
One thing
> "SJS" == Stephen John Smoogen writes:
SJS> Things don't work that quickly.
Well, they can work pretty quickly when the stars align.
SJS> We have processes that need to be followed for packages to be
SJS> reviewed and such.
I would argue that reviews aren't actually
> "RPH" == R P Herrold writes:
RPH> question / thought as to a possable extensionto the outline -- can
RPH> a 'final' revised EPEL yum repodata update be issued pointing to
RPH> the new location (and the move and later remove staged in two
RPH> pieces), so existing
> "SJS" == Stephen John Smoogen writes:
SJS> Selinux may have issues and I am trying to work through a proper
SJS> way to update the selinux policy for it without over-writing items.
You might need new policy if the new nagios does things that the old one
didn't, like call
> "PR" == Peter Robinson writes:
PR> Reminder that RHEL 5 and hence EPEL 5 has got 3 months until EOL.
PR> Packages now should really just be in bugfix and security fix only
PR> mode.
Should we be preventing the branching of new packages for EPEL5 at this
point? We
Hanging out in #epel on IRC, I've seen multiple people come by to ask
where a particular package went after it's been removed. And I just
realized that the orphaned packages report is sent only to epel-devel,
so someone subscribed just to epel-announce wouldn't see anything at
all. I don't have
> "RF" == Richard Fearn writes:
RF> I'd have thought that instead of specifying "%{_mandir}/man1/foo.1*"
RF> as the guidelines suggest, it would be safer to specify
RF> "%{_mandir}/man1/foo.1.*" to ensure it doesn't accidentally pick up
RF> an uncompressed file.
You
> "RF" == Richard Fearn writes:
RF> Thanks Jason for looking into this, and for your work on these
RF> macros in general. I take it you mean you were able to build the old
RF> version of the package (1.12-1) without issue?
Yes, it builds with and without a %clean
I have developed a workaround for the issue you reported. I am able to
build your package without issue. I have submitted the updated package
to testing and I would welcome any testing. In the meantime I will set
up another complete el5 rebuild to double-check.
Alternately, I could simply
Yeah, this is a sadly unintended side effect of the magic I had worked
out to make %clean optional. Instead it introduced situations where
%clean needed to be removed (but, of course, not every situation).
I'm still thinking about ways to handle it, but EPEL5 is fortunately not
long for this
> "d" == dani writes:
d> I think it is high time to rethink the single version of a package
d> policy, and come up with some scheme that would allow for any package
d> to maintain multiple versions in a consistent manner.
We don't have such a policy. At least Fedora
> "AL" == Avram Lubkin writes:
AL> I'm curious if anyone else has any insight. Maybe it is worth
AL> bringing up at a FPC meeting.
That would more appropriately be a topic of an EPEL meeting, since this
is purely an EPEL policy issue.
- J<
> "AL" == Avram Lubkin writes:
AL> Not needing reviews would help, but I wonder how hard it would be to
AL> make them children of python-PACKAGE. The main issue is the SRPM
AL> needs to have a different name so there is no conflict with the RHEL
AL> SRPM.
To be
> "AL" == Avram Lubkin writes:
AL> Definitely, but it runs into the same problem as 3.4 on EL7, the
AL> fact that there are few packages available and adding them when the
AL> package already exists in RHEL requires creating a separate parent
AL> package in Fedora
> "DJ" == Dave Johansen writes:
DJ> devtoolset is designed to do all of this and is already done, so it
DJ> seems that the only advantage to putting it in EPEL itself would be
DJ> to reduce the number of repos during build time.
So is devtoolset something I get
I added a set of python macros to epel-rpm-macros-6 and pushed them out
to testing. This should allow you to use the all of what's in the
regular Fedora Python guidelines as is. You will still have to %if out
some of the python3 stuff (build deps, subpackage declarations, %files,
etc.), but
Just FYI, EPEL5 and EPEL6 now have functioning %autosetup
implementations. For EPEL6 there are no caveats; for EPEL6, if you have
patches numbered higher than Patch9: then you will need to set
%el5_patches_limit appropriately. But, uh, why would you do that?
All documented at
> "DJ" == Dave Johansen writes:
DJ> Is there anything that could be done in EPEL to make the Python 3
DJ> macros be usable without requiring %{python3_pkgversion}?
Well, there has been some on and off work focused on making python
packaging less hideous. A spec
The releng folks managed to get to the bottom of this and fix the
underlying issue, so epel5 mock builds should now have the
epel-rpm-macros package available for your specfile de-crufting needs.
And remember, epel-rpm-macros-5 with functioning %autosetup is now in
testing. Should make it to
> "DJ" == Dave Johansen writes:
DJ> To my knowledge mock for EL 5 has been broken for several months
DJ> now:
I've had no problems using it. I did rebuild all of EPEL5 at least ten
times recently and while there were plenty of failures I'm pretty sure
those failures
Something went wrong when epel-rpm-macros-5 was added to the EPEL5
buildroot such that it works fine in koji but isn't actually present
when you build in mock. So if you were trying to de-cruft your specs
and found that things weren't working as you expected when you did a
mock build, that's why.
I thought it was impossible. Then I thought it was merely impossible
without some terrible hacking. But now, after a bit of inspiration in
the shower this morning, I went ahead and implemented the complete
%autosetup functionality for EPEL5. Currently built as
epel-rpm-macros-5.3 but not yet
> "SJS" == Stephen John Smoogen writes:
SJS> What bug? Sorry we really need to see some actual output and
SJS> problems here to have an idea of what we are trying to tackle.
I think he's referring to the fact that the SCL macros break if you try
to do too much with other
And I just pulled two random package with EL6 branches, changed %doc to
%license in the appropriate places, and built them in mock. Everything
came out as expected (no build failures, and the license files are in
with the rest of the documentation).
So I unfortunately have no idea what might be
> "DL" == Dave Love writes:
DL> How is the epel-rpm-macros package supposed to work? I have
DL> epel-rpm-macros-6-4 installed, which is up-to-date against
DL> epel-testing, and is supposed to make %license work like %doc but
DL> doesn't seem to have any effect.
> "TK" == Toshio Kuratomi writes:
TK> The easiest thing to do if you want a single spec file for EPEL7 and
TK> Fedora is to Requires: python-foo rather than python2-foo.
Except that FPC would like to move away from this. That's the entire
reason I've brought this up.
> "P" == Peter writes:
P> I would be worried, though, that you'll have packages that were built
P> against python that are now trying to pull in and possibly run on
P> python2 unnecessarily, and possibly detrimentally if Red Hat suddenly
P> decides to push python2
> "JH" == James Hogarth writes:
JH> Do just to be clear from your wording you are talking about RHEL
JH> python packages that are known as python-foo in RHEL rather than
JH> python2-foo there, since there is no other python within base to
JH> cause confusion?
Right,
One annoying difference between packaging for Fedora and EPEL7 (and
probably older) is the fact that Python packages in Fedora are required
to provide "python2-foo" whereas many EL7 packages don't. This leads to
ifdefs and unpleasantness, and kind of complicates our ability to hide
some details
> "CD" == Christian Dersch writes:
CD> [...] for EPEL 6 I'm
CD> unable to apply security fixes as one of them (the most important
CD> one...) requires C++11 features not available in gcc 4.4 :( What is
CD> the best strategy for such a package?
Maybe ask someone who is
It appears that it's actually trivial to add %autosetup to RHEL6. You
just need the Fedora macros plus a couple of extra definitions. That's
in git now.
I don't want to push this to testing because there's another version
soaking in testing which I don't want to supersede. However, you can
I should also note that epel-rpm-macros-6-2 has gone to stable, and it
adds the requested node.js macro.
- J<
___
epel-devel mailing list
epel-devel@lists.fedoraproject.org
http://lists.fedoraproject.org/admin/lists/epel-devel@lists.fedoraproject.org
The epel-rpm-macros-5-1 package for EPEL5 should now be in stable but I
have not yet requested that it be added to the buildroot. This adds a
number of things which I've mentioned earlier (Group:, BuildRoot: and
%clean not needed; buildroot automatically cleaned in %install) and
allows packagers
> "o" == opensource writes:
o> The following packages are orphaned and will be retired when they are
o> orphaned for six weeks, [...]
o> directfb orphan, kwizart, thias 17 weeks ago
How does the retiring actually happen? I know it's a manual process,
> "RF" == Richard Fearn writes:
RF> My point was that you couldn't get it to build because there was no
RF> libewf to build it against. There is now, because the
RF> disktype/libewf update has finally been pushed.
Cool, then I suppose it will all be cleaned up after
> "PH" == Paul Howarth writes:
PH> does not. It's probably still there because people can't remember
PH> whether it was EL-5 or EL-6 that removed the need for it, and left
PH> it there to be on the safe side.
That's good to know. I also think there's more than a fair
> "DJ" == Dave Johansen writes:
DJ> And it's not helped by the fact that the version of rpmlint on EL 5
DJ> and 6 warns when it's missing.
Interesting. Well, we can fix rpmlint. And I can grep at least the
rawhide specfiles for remaining instances, generate a
> "DC" == Dan Callaghan writes:
DC> All of these broken dependencies onto python-setuptools-devel seemed
DC> a little strange to me so I dug into it.
Thanks for doing this. I thought it was a bit odd but figured it was
better to actually get the report out there rather
I'm now working on some magic macros for EPEL5. Currently (on my
machine, at least) you can use %license and don't need BuildRoot:. I'm
curious about some other boilerplate constructs, though.
%defattr in %files:
I've been told that even EPEL5 doesn't need this, but still it
persists. Can
> "DG" == Dennis Gilmore writes:
DG> When you are ready for the change to be made, please file a ticket
DG> with releng, we will need to coordinate a koji and comps change.
Yep, that was part of my plan. After committing the tiny fixes needed
for those few packages (many
I performed a mass build of EPEL5, skipping only a few packages that
take absolutely forever to build (libguestfs, thunderbird-lightning,
ikarus and pypy). I also skipped rubygem-eventmachine because its test
suite hangs forever. Turns out that I should also have skipped
I've now done many mass rebuilds both with and without epel-rpm-macros
in the buildroot and have found exactly 31 failures which result from
the presence of the macro package. This macro package enables, in
EPEL6, the use of %license in the %files section of the spec without
having to
> "TK" == Toshio Kuratomi writes:
TK> We would have been in a lot better place today if we had separate
TK> packaging of python2 and python3 packages in Fedora so that they
TK> were never in sync there but that's not something we can probably
TK> change now
Nothing
> "DF" == Denis Fateyev writes:
DF> If we just could work "the same SRPMS name" problem around ;-)
DF> Healthy repos with the master branch orphaned [1] may look a little
DF> weird to users...
That is not abnormal for EPEL-only packages, though. The dead.package
file in
> "AT" == Antonio Trande writes:
AT> %{__global_ldflags} is another macro that it's not available in
AT> EPEL6.
It appears that gets exported as a shell variable in the scope of %build
by %__build_pre, as well as being mentioned in the %cmake, %cmake_kde4,
%qmake_qt4
> "JLT" == Jason L Tibbitts writes:
JLT> Does that actually work on EL6?
Looking at the spec, it seems obvious that it works on EL6. The package
isn't branched for EL5 but if anyone knows if -Wl,-z,relro will work
there than I'll add it there as well (once I get around
Lately I've been working on an EL6 branch of epel-rpm-macros with the
goal of removing the need for some of the %ifdefs and line noise and such
required if
you want to have one spec which builds on Fedora and EL6.
Right now it simply enables %license in the %files section (mapped to
%doc as the
> "AT" == Antonio Trande writes:
AT> This is not current tktable release.
Well, it's what is currently on the master mirror. I'm not building
from git; I'm taking the current set of released SRPMs. I could take
them from testing instead if that's thought to be
> "DC" == Dan Callaghan writes:
DC> Would it be easier to request the RHEL packages to add a virtual
DC> Provides for the python2-* name? That is, python-setuptools in RHEL
DC> could provide python2-setuptools.
I can't imagine Red Hat going through all of the
74 matches
Mail list logo