Hi Ricardo, Michael, Marcin, and Brendan,
On Wed, 2011-11-30 at 21:16 -0200, Ricardo Salveti wrote:
On Tue, Nov 29, 2011 at 11:17 PM, Michael Hope michael.hope@linaro.org wrote:
We don't have a libc as there hasn't been the need. The cross compiler could either target the Fedora ARM port or the Linaro LEBs.
I believe it could simply target the Fedora ARM port, but then don't know if it'll be easily compatible with other rpm-based distros.
This is a good opportunity to ensure that it is the case :) If there is diversity at the fundamental level sufficient to break compatibility between RPM distributions we have a lot more work to do on cross-distro compatibility (and I want RPM/non-RPM compatibility to be great also). We would certainly welcome data about how compatible we are today.
Well, guess making the Linaro native package available should help you understanding if there's any need to do the switch. Even if not entirely switching to the Linaro GCC, you could simply make the package available for the people to try and help tracking and comparing bugs and issues with the native distro GCC.
That is reasonable, us switching to anything other than the official Fedora GCC is not a reasonable outcome for the Fedora ARM project :) In my mind, I enjoy that we have GCC maintainers working for Red Hat who maintain both our commercial and Fedora GCC packages. I welcome and support the work Linaro are doing, but I think it would not be to the greater benefit of our distribution to outsource the toolchain itself.
Once packages are in rpm format it's very straight-forward for anybody to start using them, pulling updates, etc.
I'd have to bring that up with management. We'll support you if you use it but producing and maintaining the packaging is an overhead.
This is solved at our LEBs by enabling the packaging recipes at Launchpad, that then merges the packaging souce with the Linaro GCC trunk and push the package to be automatically built at our PPAs. If you have some sort of a similar system, that could trigger new packages and rebuilds once the Linaro GCC bzr tree gets updated, I believe we can at least help setting up the environment and building the first packages.
We do have a capability to have personal packages and our build system can import sources from various places. Having said that, I do not believe a fully automated build based on polling your trees is going to work with the current technology. It would, though, be fairly low overhead to have a "package" in Fedora that is really just an RPM spec file and a clone of the Linaro sources you/we can pull into frequently.
To have proper maintenance and such would then be a quite overhead, and would need an agreement with the management.
I'm not convinced we need to have a "product" grade toolchain yet. What we could start with is just a period sync of the latest Linaro toolchain into a Fedora ARM package and see what the interest is in developing it.
As to conference calls, I'm flexible on the timing myself.
Thanks,
Jon.