Set tested distros list
cover Set tested distros list
I had hoped to be able to have a patch setting this list much earlier, but for several reasons this did not happen. The list is quite short, but it's currently all we have on record.
(Note, I excluded Fedora 16 as that isn't even at final RC stage yet.)
The following changes since commit e3c5d7a90ae179af0ac128ea4dfc1a5a6602c2e4:
libtool: Fix an issue where unnecessary rpaths were being injected (2011-09-26 17:56:44 +0100)
are available in the git repository at:
git://git.pokylinux.org/poky-contrib paule/sanity-distros
http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=paule/sanity-distros
Paul Eggleton (1):
conf/distro/poky.conf: set SANITY_TESTED_DISTROS
meta-yocto/conf/distro/poky.conf | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-)
Signature · 1 line
1.7.4.1
PATCH 1/1 conf/distro/poky.conf: set SANITY_TESTED_DISTROS
Specify a list of tested distributions, based on those tested prior to the Yocto Project 1.1 release as recorded on the wiki.
--- meta-yocto/conf/distro/poky.conf | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-)
meta-yocto/conf/distro/poky.conf +12 -0
@@ -61,3 +61,15 @@ CONNECTIVITY_CHECK_URIS ?= "git://git.yoctoproject.org/yocto-firewall-test;proto https://eula-downloads.yoctoproject.org/index.php \ http://bugzilla.yoctoproject.org/report.cgi" +SANITY_TESTED_DISTROS = " \ + Ubuntu 10.04 \n \ + Ubuntu 10.04.3 LTS \n \ + Ubuntu 11.04 \n \ + Fedora release 13 (Goddard) \n \ + Fedora release 14 (Laughlin) \n \ + Fedora release 15 (Lovelock) \n \ + openSUSE 11.2 \n \ + openSUSE 11.3 \n \ + openSUSE 11.4 \n \ + " +
Signature · 1 line
1.7.4.1
Re: [PATCH 1/1] conf/distro/poky.conf: set SANITY_TESTED_DISTROS
In reply to a message not in the archiveHi Matthew,
Quoted 6 lines · On Tuesday 27 September 2011 16:41:42 McClintock Matthew-B29882 wrote:
Quoted 2 lines · <[email protected]> wrote:
Specify a list of tested distributions, based on those tested prior to the Yocto Project 1.1 release as recorded on the wiki.
What are the specific qualifications to make this list? I have run on CentOS 5.X in addition to ones on this list.
Well, there are two aspects:
1) We need the concrete information on what was tested; we've been collecting this on the following wiki page:
https://wiki.yoctoproject.org/wiki/Distribution_Support
So, I'm not the person that sets policy on this (I'm not sure we have a policy yet) but thinking off the top of my head, ideally a Poky build of core-image- sato should run "out of the box" (after installation of necessary packages, that is) with no hacks, although if we are able to document any necessary changes clearly I suspect even that shouldn't be an issue.
2) More practically we need to know what the ID string is for the distro. The code in sanity.bbclass looks at /etc/redhat-release, /etc/SuSE-release, /etc/lsb-release etc. to determine this.
FWIW I agree that CentOS should be on the list but I do not have a system running it to test, and I was missing the above information when I put together the patch. If you're confident that it works well for builds and can help with the above then I'm sure we can add it.
Cheers,
Paul
Signature · 2 lines
Paul Eggleton Intel Open Source Technology Centre
Re: [PATCH 1/1] conf/distro/poky.conf: set SANITY_TESTED_DISTROS
In reply to a message not in the archiveQuoted 4 lines · On Tuesday 27 September 2011 17:19:50 you wrote:
The other question is what MACHINE's we build for - it should be fairly comprehensive to just test for one, but it's hardly adequate... I'd assume x86, arm, ppc, mips all need to be build tested and boot tested?
I'm not sure we need to go that far. We do run such comprehensive tests for a small list of common distros as part of the core testing done by the Yocto QA team. However, whilst it's entirely possible that a host distribution might cause failures with one MACHINE / TARGET_ARCH and not another, in practice I would expect that if a build of a fairly comprehensive image (such as core- image-sato) for one machine is able to complete successfully and run within QEMU then that should be a good indicator that the host distro is suitable. Additionally, if builds are being run on that system on a regular basis (e.g. the autobuilders, or a developer's main machine) then that would of course increase the level of confidence in that distribution.
Naturally, resources allowing we would like to have full batches of tests run on more distributions, but I don't think it's practical to commit to testing every distro.
Jiajun might have some other opinions/experience here, Jiajun...?
Cheers,
Paul
Signature · 2 lines
Paul Eggleton Intel Open Source Technology Centre