On 11/13/2010 02:17 PM, Karanbir Singh wrote:
On 11/13/2010 08:25 AM, Farkas Levente wrote:
/ - what os, version do you use on build hosts?
/>
CentOS-5.5
/ - which version of mock and ccache do you use?
/
no ccache, mock-0.6.3
/ - and what mock config do you use?
/
nothing interesting in the mock configs.
If there is nothing interesting in the mock configs why not share them. The config really defines the build environment not the host and not necessarily the version of mock. It is the default installs + the packages to meet dependancies and where they come from.
/ since it's well known that there are some packages (like bash, nss)
/ ? because you have those problems, does not mean everyone else will.
/ which can not be build in mock on rhel-6:
/>>/ https://bugzilla.redhat.com/show_bug.cgi?id=613392 />>/ https://bugzilla.redhat.com/show_bug.cgi?id=609201 />>/ so it's rather strange that build complete without issues. /
Speak to the people who have the problems ? These packages have all clearly built not only for fedora, but also for red hat - and going by the changelogs, quite a few times.
- KB
Since I have built all but 36 packages on both 1686 and x86_64 with rpmbuild I am very interested in resolving the differences in build environments. Once I too have a stable build environment I would be interested in working on the rebranding etc. But I would like to be able to test and make sure my changes are not causing the problems not the Build environment. I am building on RHEL6 + updates.
Hubert
On 11/29/2010 04:36 AM, Hubert Bahr wrote:
I am very interested in resolving the differences in build environments. Once I too have a stable build environment I would be interested in working on the rebranding etc. But I would like to be able to test and make sure my changes are
its the sources that need to be audited, not the binaries.
besides, don't use static mock roots anyway in CentOS, never have. Don't have the luxury of that level of simplicity
once we have a bit of breathing time, I'll get all the code that is being used here published. Till then, its focus on getting release.
- KB
Karanbir Singh wrote:
On 11/29/2010 04:36 AM, Hubert Bahr wrote:
I am very interested in resolving the differences in build environments. Once I too have a stable build environment I would be interested in working on the rebranding etc. But I would like to be able to test and make sure my changes are
its the sources that need to be audited, not the binaries.
besides, don't use static mock roots anyway in CentOS, never have. Don't have the luxury of that level of simplicity
Thanks, That gives me more insight into the process. Non static mock root implies that changes are necessary to do a complete build. Hubert
once we have a bit of breathing time, I'll get all the code that is being used here published. Till then, its focus on getting release.
- KB
CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
On 11/29/2010 04:34 PM, Hubert Bahr wrote:
besides, don't use static mock roots anyway in CentOS, never have. Don't have the luxury of that level of simplicity
Thanks, That gives me more insight into the process. Non static mock root implies that changes are necessary to do a complete build.
With EL6 there have been some really strange build roots as well, we might need to make a pragmatic call at some stage on how-to work with or without those. One of the most odd ones is https://bugzilla.redhat.com/show_bug.cgi?id=644778
where buildroots even for the same package were different on different architectures; in this case, it 'feels' like its because of the rolling builds where queues ran independently for different architectures. And there is no clean way of working around this, unless you don't mind cross builds waiting for a sync 'pulse' for each package.
Can be mitigated to some extent by running packages-as-tasks rather than builds-as-tasks; but then were now talking about spiritual issues subject to wildly divergent interpretation.
- KB
Hi Karanbir,
On Thu, 2010-12-02 at 02:18 +0000, Karanbir Singh wrote:
With EL6 there have been some really strange build roots as well, we might need to make a pragmatic call at some stage on how-to work with or without those. One of the most odd ones is https://bugzilla.redhat.com/show_bug.cgi?id=644778
That one looks very similar to the one I reported here in the thread about file conflicts from an x86_64 and i686 build I did.
That issue turned out to be caused by the fact that in my base system I did not have redhat-rpm-config installed and in the mock build root it did get pulled in, so in the latter %_binary_filedigest_algorithm 8 got set. My conclusion was that rpm can't compare the resulting doc and config files as being equal as they use different checksum mechanisms.
Regards, Leonard.
On Fri, 2010-12-03 at 00:16 +0100, Leonard den Ottolander wrote:
That one looks very similar to the one I reported here in the thread about file conflicts from an x86_64 and i686 build I did.
Never mind. Same errors, different cause.
Regards, Leonard.
-
Hubert Bahr -
Karanbir Singh -
Leonard den Ottolander