[scl.org] Bootstrapping devtoolset?
Is there a method for bootstrapping devtoolset (-3, -4)? I used to be able to build everything in a mock for devtoolset-2, but for the newer ones I find that they seem to have cyclic build dependencies where most builds want to pull in everything that would only be available after the successful build (currently stuck at rh-java-common). How was the bootstrapping done? ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Bootstrapping devtoolset?
On Tue, Dec 6, 2016 at 12:05 PM, Mikolaj Izdebski wrote: > Java collections were typically bootstrapped by using older packages > from other collections or from RHEL. Specifically, rh-java-common was > built using maven30 RPMs and maven30 was built using RHEL 7 content. > > CentOS bootstraps SCLs by using external repositories with prebuilt > RPMs, like this one: > http://cbs.centos.org/koji/externalrepoinfo?extrepoID=11 Unfortunately I cannot see what’s inside this repository or some siblings I've tried (access denied). Does it basically contain the same packages as the final sclo repo or are there some hand-tweaked packages aimed at getting through a bootstrap? I guess that bootstrapping was a manual process with lots of dirty hacks and workarounds ;). Is this a good impression of how it went? ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Bootstrapping devtoolset?
On Tue, Dec 6, 2016 at 11:54 AM, Stefan Ring wrote: > Is there a method for bootstrapping devtoolset (-3, -4)? I used to be > able to build everything in a mock for devtoolset-2, but for the newer > ones I find that they seem to have cyclic build dependencies where > most builds want to pull in everything that would only be available > after the successful build (currently stuck at rh-java-common). After having gone through all the trouble of rebuilding rh-java-common and maven30, I noticed that there is a set of devtoolset-6-* packages now which do not depend on any of that Java cruft and are straightforward to build :). ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] rh-python36-build alternative in rhel
On Wed, Nov 22, 2017 at 2:45 PM, Jiri Novak wrote:
> Hello,
>
> I'm currently trying to port python app to use rh-python36 SCL (built
> using mock+rpmbuild), I've managed to make it build correctly in the scl
> paths (and using python3 for the packages that need it during build) if
> I install rh-python36-build from centos to the mock root.
> Is there a way to do this with rhel packages too?
Do what?
> The issue is that
> without it,
Without what?
> %{scl} in specfile is not populated, with it, it's properly
> set to rh-python36. Should I pass the value in different manner than in
> centos? or what might I be doing wrong?
I have no idea what you are trying to achieve.
___
SCLorg mailing list
SCLorg@redhat.com
https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] rh-python36-build alternative in rhel
On Tue, Nov 28, 2017 at 1:47 PM, Jiri Novak wrote:
> Hi,
>
> according to this (
> https://developers.redhat.com/blog/2014/12/04/add-packages-to-python-2-7-software-collection/
> ), I'm trying to create my own extension of the rh-python36, though
> doesn't build:
>
> + '[' '%{buildarch}' = noarch ']'
> + QA_CHECK_RPATHS=1
> + case "${QA_CHECK_RPATHS:-}" in
> + /usr/lib/rpm/check-rpaths
> + /usr/lib/rpm/check-buildroot
> + '%{rh-python36_os_install_post}'
> /var/tmp/rpm-tmp.k3bzOV: line 316: fg: no job control
> error: Bad exit status from /var/tmp/rpm-tmp.k3bzOV (%install)
>
> most probably because the variables are not expanded?
> What did I miss? Specfile for refenrence in attachement.
What is this spec file? It looks awfully complicated.
I'm getting you want to build something against rh-python36. Why don't
you take an existing package, let’s say, rh-python36-python-six, and
adapt it?
___
SCLorg mailing list
SCLorg@redhat.com
https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] rh-python36-build alternative in rhel
On Tue, Nov 28, 2017 at 2:40 PM, Jiri Novak wrote: > Hi > > I need to add more libraries to rh-python36, which would require using > rh-python36-build package and I've been told it is wring and I should do > my own SCL. > > then I need to make the application use the extended SCL with added > libraries. > > Just running it agains rh-python36 is not enough because of missing > dependencies :( > > The specfile should create packages test-python36-build and > test-python36-runtime as a collection that extends rh-python36 Is this even possible? If your collection is called test-python36, it will live in another directory hierarchy than rh-python36 and will not be able to access the packages from there. I think you're in for a world of pain. You can of course start a new collection by renaming everything in the rh-python36 package sources, picking up just the packages you need, thus creating a new, derived collection. ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] rh-python36-build alternative in rhel
> What else is the right way to make RPM adding more libraries available > when in scl enable rh-python36 bash? Create your own mirror of rh-python36 and just add your packages to it? ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] rh-python36-build alternative in rhel
On Wed, Nov 29, 2017 at 11:12 AM, Jiri Novak wrote: > Stefan Ring wrote: >>> What else is the right way to make RPM adding more libraries available >>> when in scl enable rh-python36 bash? >> Create your own mirror of rh-python36 and just add your packages to it? > Sounds like dirty hack as rh-python36-build was intentionally not > shipped in rhel. or by mirror you mean to rename it to xxx-python36 and > rebuild all packages I need? No, I mean a real mirror that carries all the original packages, to which you can also add your own. ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Rebuilding sclo packages in mock
On Fri, Oct 4, 2019 at 4:26 PM isdtor wrote: > > Hi all, > > I have a problem similar or at least related to > https://www.redhat.com/archives/sclorg/2016-May/msg00020.html > except I'm trying to do it all in mock. > > There's a custom config file with > > config_opts['chroot_setup_cmd'] = 'install @buildsys scl-utils-build > httpd24-build shadow-utils' > > and the command I'm running to just rebuild this package is > > mock -r scl-6-x86_64 -D "scl httpd24" httpd24-1.1-18.el6.src.rpm No idea, works for me on CentOS 6 (mock-1.2.21-1.el6). ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] How to use SCL inside a container
On Mon, Jul 27, 2020 at 12:23 PM Jonathan MERCIER wrote:
>
> Dear I tried to build a container both with docker and buildah and got the
> same message error from gitlab-ci
>
> ---> https://gitlab.com/jonathan-dlang/scl_metapackage/-/jobs/657306213
No access rights.
> $ scl load "${SCL_VENDOR}-${SCL_COLLECTION}${DLANG_VERSION}"
> 31Missing function scl in your environment!!!
>
> What is those scl functions ?
>
> We can see too that while scl_source is calld (line 28) the list of loaded
> collection is empty (line 29)
Do you know how to use scl outside of a container? Because it does not
actually make a difference.
> I tried to use a scl_enable file as it is done by sclorg ->
> https://github.com/sclorg/mariadb-container/blob/master/root-common/usr/share/container-scripts/mysql/scl_enable
>
> with the same error
If it’s only one line in the CI recipe, just use scl enable {scl_name}
-- command (line).
___
SCLorg mailing list
SCLorg@redhat.com
https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] devtoolset-6 and stdlibc++
On Mon, May 10, 2021 at 6:25 PM Matthew Wheaton wrote: > > > Hi, > > This can't be the first time this has come up. I was very pleased to > discover that I could install the gcc-6.3 compiler on a rhel6 host using > devtoolset-6, but now discover that the accompanying stdlibc++ is not > included. > > I provide shared libraries as part of an API for my work and it would be > greatly beneficial to provide a runtime environment that includes the > libraries that come with gcc-6.3. Unless the customer can use a rhel8 > equivalent, this is the best way to provide C++11 functionality that I know > of. > > Is there a way to provide the appropriate stdlibc++? The binaries produced by devtoolset are supposed to run against the system runtime (libstdc++), so there is nothing else to be shipped. It sounds like you would want to just build gcc yourself (it’s easy!) Then you will need to ship the runtime with the binaries produced by it. ___ SCLorg mailing list SCLorg@redhat.com https://listman.redhat.com/mailman/listinfo/sclorg
