On Aug 16, 2008, at 11:29 AM, Martin MC Brown wrote: > And to pick up on your main point here, significant changes dont > have to mean incompatible ones.
Indeed! The focus on the taxonomies is ultimately on the incompatible changes. New features don't annoy customers (at least, very rarely! ;-). The terms Major, Minor, Micro are not judgments on how much work went into a release or the importance of the release to the user community, they are about what kind of incompatible change is permitted. Anyway, some background reading: http://opensolaris.org/os/community/arc/policies/release-taxonomy/ http://opensolaris.org/os/community/arc/policies/interface-taxonomy/ > I'd say you were over-specifying the semantics of the situation :) Well I'm reviewing an ARC draft. Sorry I don't know how familiar you are with ARC review process, but it can be rather formal at times [some might say that's an understatement] (and yet sometimes it'll be very superficial). It's always best to cross all t's and dot all i's in the materials before starting the case. If your team would like me to sponsor this project as a Fast Track, then I'm required to put my ARC hat on and pick all the nits now, before the case goes to the committee. While we can't predict which topics other ARC members might need more info on, my job becomes to predict most of them and make sure the spec addresses them upfront. > I'll leave the debate about whether the probable reviewers are not > on this list for another time :) Sorry I don't follow the joke, but anyway it's not a debate, just a statement on the [lack of] membership overlap between LSARC and webstack-discuss. Anyway, thanks for clarifying and reinforcing that 5.1 is an upgrade to 5.0. Customers are expected to migrate in short order (not instantaneously, but in short order) and 5.0 soon goes away instead of being maintained in parallel. Therefore, 5.1 is not an all-new application so the ARC spec needs to highlight what is different from 5.0 so the review can focus only on that part. > I don't see any need to go into more detail in the ARC case, most > end-users are unlikely to use the ARC case as their only source for > upgrade information. Upgrades, along with interface changes, are areas which receive most attention at an ARC review. No, it is not a user-centric doc and we don't expect users to read it as their source of info. It only needs to have enough info to convince ARC the project team has thought through all the upgrade scenarios and what these scenarios will impose on users. [The following is not an ARC requirement, so I remove my ARC hat and put on my Web Stack lead hat] > YMMV on which Linux installs it into such a global location, not all > of them do, and not all of them follow even Debian is what I'm using. Out of the box usability is one of "the" key drivers for Web Stack project and not having something as commonly used as the mysql CLI available out of the box after installing the MySQL package (nobody has /usr/mysql/$VERSION/bin in their PATH out of the box) is lame usability. We really must seek to do better than this. If not all Linux distros do this, to me that is an opportunity to do better than some of the Linux distros (and at least equal to Debian). It shouldn't be a justification to be equally user-unfriendly. Can OpenSolaris be the best? I'd like to see that...
