My test / development host runs CentOS-5.x x86_64. There I have successfully created an rpm for GiT-1.6.1-1 for x86_64. Now I would like to build the same rpm package for CentOS-5.x i386, but on the x86_64 machine.
Is this even possible? If so, are there any resources that someone can refer me too that explains how?
James B. Byrne wrote:
My test / development host runs CentOS-5.x x86_64. There I have successfully created an rpm for GiT-1.6.1-1 for x86_64. Now I would like to build the same rpm package for CentOS-5.x i386, but on the x86_64 machine.
Is this even possible? If so, are there any resources that someone can refer me too that explains how?
Yes. In fact, I suspect the i386 rpm's for CentOS are built on x86_64
The easy way -
rpmbuild --target="i386" foo.spec
You will need all the 32-bit devel libraries.
The proper way would be to install mock (the version in EPEL is best) and use that.
On 03/31/09 09:44, James B. Byrne enlightened us:
My test / development host runs CentOS-5.x x86_64. There I have successfully created an rpm for GiT-1.6.1-1 for x86_64. Now I would like to build the same rpm package for CentOS-5.x i386, but on the x86_64 machine.
Is this even possible? If so, are there any resources that someone can refer me too that explains how?
https://fedoraproject.org/wiki/Projects/Mock
On Tue, Mar 31, 2009, James B. Byrne wrote:
My test / development host runs CentOS-5.x x86_64. There I have successfully created an rpm for GiT-1.6.1-1 for x86_64. Now I would like to build the same rpm package for CentOS-5.x i386, but on the x86_64 machine.
Is this even possible? If so, are there any resources that someone can refer me too that explains how?
We do this sort of thing using the free VMware server with VMs for the appropriate 32-bit Linux, Solaris, FreeBSD, etc. VMs are particularly useful in development as it is very easy to take a snapshot, experiment, and revert back to the snapshot if things go Horribly Wrong(tm).
Bill
Thanks for the pointers on cross building rpms. I have downloaded mock from epel and will take a look at that shortly.
For the moment however, I am working on a straight x86_64 build of ruby-1.9.1-1 and I am running into this error:
checking for nroff... /usr/bin/nroff creating config.h configure: creating ./config.status config.status: creating Makefile + %make /var/tmp/rpm-tmp.57843: line 58: fg: no job control error: Bad exit status from /var/tmp/rpm-tmp.57843 (%build)
I cannot fathom what the foreground call is supposed to accomplish or why it is happening.
Can anyone nudge me towards an online reference where I might discover how to interpret this?
On Tue, March 31, 2009 12:57, James B. Byrne wrote:
For the moment however, I am working on a straight x86_64 build of ruby-1.9.1-1 and I am running into this error:
I spoke too soon. I must have a system configuration problem because when I returned to git and tried the easy cross build:
$ rpmbuild --target="i386" git.spec
I got a very similar error:
Building target platforms: i386 Building for target i386 sh: line 0: fg: no job control
So, what am I supposed to do with respect to fg and job control?
On Tue, March 31, 2009 13:00, James B. Byrne wrote:
So, what am I supposed to do with respect to fg and job control?
Found it. No etc/at.allow and no etc/at.deny means only root can submit jobs.
On Tue, Mar 31, 2009 at 12:57 PM, James B. Byrne byrnejb@harte-lyne.ca wrote:
/var/tmp/rpm-tmp.57843: line 58: fg: no job control error: Bad exit status from /var/tmp/rpm-tmp.57843 (%build)
I've seen this error most often when using rpm specs built for a different system (e.g., Mandriva or Suse). It's often caused by a missing macro definition in the specfile.
Can you post the specfile?
-
Bill Campbell -
James B. Byrne -
Kwan Lowe -
Matthew Hyclak -
Michael A. Peters