On Nov 5, 2010, at 3:30 PM, Sriram Narayanan wrote:
> On Sat, Nov 6, 2010 at 12:45 AM, Jeff Johnson <email@example.com> wrote:
>>> We're ready to move forward :)
>> OK. I'll start-up the OpenIndiana VM buildbot again. Note that
>> the issue with running the OpenIndiana VM is largely
>> that its running on a devel box here at home, and
>> I'm a bit short on memory and cycles to run the number of VM's
>> that I'm attempting.
> Ok, I'll see if we can set up a buildbot at one of our user's
> computers, then. All Belenix work happens on our home desktop
> computers :)
I sorted through the CVS issues on the OpenIndiana VM I was seeing before.
Reverting from cvs-1.12.x to cvs-1.11.23 (@rpm5.org uses cvs-1.11.22) fixed
the issue for me.
OpenIndiana now has a "green waterfall" and all build logs attached at
The tests are still a bit flimsy, yes, but there's likely coverage at ~60%
of all RPM code, and closer to ~80% on the primary build/install/query/erase
paths. I can regenerate LCOV/GCOV logs for rpm-5.3.6 if really interested;
last I looked (fair number of changes since rpm-5.3.1) is available here
There are a few issues with linker relocations seen while building
popt, but you are almost certain to encounter those same issues,
I had no problem finding other instances of the same problem(s)
searching with Google.
I still need to build/install/link elfutils for RPM, but most everything
else Just Works afaict. I've already typed
and I could easily use rpm-5.3.6 to build packages. Again
getting ELF internal to RPM does a better job than external
helpers for library sonames. And its not too hard, just I
If you want to try a buildbot, there's a "INDIANA_rpm_release"
slot that you could plug into on
anytime you want (I'll send a password privately if/when you are ready).
I'd like to see a SPARC buildbot somehow: the SPARC architecture
exposes certain bugs (like alignment) faster/better than
any other architecture.
73 de Jeff
Received on Mon Nov 8 16:03:34 2010