On Tue, Jan 1, 2013 at 12:58 PM, janI <j...@apache.org> wrote: > On 1 January 2013 18:50, Rob Weir <robw...@apache.org> wrote: > >> On Tue, Jan 1, 2013 at 12:06 PM, janI <j...@apache.org> wrote: >> > +1 to your definition of "supported", it is funny I just had somewhat the >> > same discussion today. >> > >> > Regarding lifecycle, I would like to suggest that we only support the >> > latest release, otherwise we stretch our resources pretty thin. We can >> of >> > course have a statement that we in general will have a look at critical >> > bugs, but they will only be solved in the latest release. >> > >> >> And thinking a little further, there might be something between >> "supported" and "deprecated", or at least there might be different >> levels of confidence we might have. >> >> For example, I don't think we're doing any testing with Widows Vista. >> We tested Windows XP, 7 and preview version of 8. We have limited >> resources. >> >> So is Vista supported? It certainly isn't deprecated. But neither >> is it getting the full QA treatment. Similar questions for Linux >> releases. We don't test every release of every distro. We pick the >> major ones, such as the Ubuntu LTS releases. >> >> One way of handling this could be: >> >> 1) Define our "Class A" platforms, the ones that we give the full test >> attention to. Similar to how we treat translations, this list can >> grow given sufficient volunteers to cover the testing, and (if bugs >> are found) the fixes. >> >> 2) Class A platforms (or "primary platforms" or "tested platforms" or >> "supported platforms" -- whatever we call them) are the ones we >> encourage users to use. >> >> 3) For other platforms we make a wiki-page per platform, were we can >> track notes from users on an unique issues they find on that platform. >> These combinations are not supported, but may often work. But we >> make it easy to collect observations about AOO on that platform, and >> make it easy for users to find that info. >> >> If we do this, then our support statement could read something like: >> "We have tested and qualified Apache OpenOffice X.Y on the following >> operating system versions. Other operating system versions may work >> as well, but may require additional configuration. For the latest >> information please consult the following wiki page..." >> >> -Rob >> >> I do not know this, but would it be possible to make a QA package (script > or something) that would make it easy for skilled users to do QA of a > platform (e.g. vista). I have f.x. vista running and could do it, but I do > not have a clue what should be done, and there could be other users like me > out there. >
We have some automation, but not enough to fully test a release. It is more at the level of checking a build to make sure it did not have gross defects, so more of a "smoke test". Most of the testing is manual. So if we had a model where a set of volunteers wanted to bump a new platform up to a "fully supported" platform, then we would have a set of automated and manual tests that the volunteers would need to run. -Rob > Microsoft have something I think they call certification scripts, that > checks if your platform is ok for a given product, could we do something > the same, that would be a one-time effort. > > Jan I. > > >> > rgds >> > Jan I. >> > >> > >> > On 1 January 2013 17:59, Rob Weir <robw...@apache.org> wrote: >> > >> >> When a commercial software vendor says a configuration is "supported" >> >> it means something, typically that to the extent the software license >> >> includes an entitlement to support, that the vendor will provide that >> >> service for that configuration. So saying something is "supported" is >> >> essentially an obligation. >> >> >> >> With a volunteer-run, open source project, "supported" cannot mean >> >> quite the same thing. We're not obligated, in any contractual sense, >> >> to provide anyone with anything. That's the nature of a volunteer >> >> effort. >> >> >> >> However, users and organizations considering OpenOffice will naturally >> >> think in terms of "support", even if they user that term loosely. We >> >> use that term as well, in our release notes, etc. But I think we >> >> ought to have a more precise definition of what we mean when we say >> >> something is "supported", in order to avoid any confusion. This >> >> question has come up recently, with regards to the status of Windows >> >> 8, where that OS had not been released at the time AOO 3.4.1 was >> >> released. >> >> >> >> So here's a strawman proposal for what "supported" means for us. >> >> >> >> 1) "Supported" is a statement we make about a specific version of AOO >> >> used with a specific platform, e.g., AOO 3.4.1 with Windows XP SP3 or >> >> AOO 3.4 with Ubuntu 12.04 LTS. >> >> >> >> 2) "Supported" means we encourage use of AOO in that configuration. >> >> We have high confidence that the combination is stable, that it works >> >> well and is safe. >> >> >> >> 3) Our confidence in stating something is supported should have a >> >> solid basis in testing. Something is not "supported" by us guessing >> >> it should work. It is supported only after we have successfully >> >> completed testing of that release with that platform. We probably >> >> should define exactly what level of testing is required. >> >> >> >> 4) "Supported" also implies that the supported configuration is >> >> sufficiently available and there is sufficient expertise that we have >> >> confidence that users will have a high quality experience seeking >> >> support on the forums and user list. >> >> >> >> 5) "Supported" also implies that we stand behind that release and will >> >> take necessary steps to correct *critical* bugs, especially security >> >> flaws, via rapidly produced point releases where necessary. >> >> >> >> Note that these are all expectations that a user might have, though >> >> any given user might think that "supported" means only a subset of >> >> these. >> >> >> >> What we probably really need is more of a lifecycle statement, >> >> including when support for a configuration ends. >> >> >> >> -Rob >> >> >>
