On Thu, Nov 27, 2008 at 1:36 AM, Henning P. Schmiedehausen < [EMAIL PROTECTED]> wrote:
> Ian Boston <[EMAIL PROTECTED]> writes: > > >Any comments ? > > Folks, I might be late to the show (the whole discussion seems to be > happening on Nov 20th inside a few hours [1]), but please: > > This versioning scheme (along the spec) is bad. Shindig will be > considered "not ready" by not hitting a 1.x version. > > By that arguing, the Apache Webserver would be at > 1.1-<some-large-number>. What if you want to rearchitect the internal > code of Shindig and need to reflect that in a version number? > 0.8.1-2.1 ? that was also one of my concerns. there's 2 ways of thinking here: 1) We'll get lots of questions about "What version of OpenSocial does this support" 2) It's confusing to have a OpenSocial spec version number as shindig version number, and < 1.0 sends the wrong signals The problem is that unless you want to go for the unwieldy long and confusing version numbering (1.0.0-0.8.1 or 0.8.1-1.0.0) you can't really address them both ... alternatives are to go for a versioned 'spec support' (the standing proposal) aka 0.8.1-REV, or rely on documentation (and common sense) to communicate which spec level is supported. I say common sense, and that is why i *am* persuadable to this 1.0.0 argumentation, because shindig *always* is somewhere between and often even a combination of [Finishing latest spec support], [Latest release spec support] and [Future spec support], which is hard to reflect in version numbers (0.8.1-0.9.partial-0.10pre ?), and having a versioning that de-couples this meaning saves us this pain. One of the most frequent questions on the list in the future will be > "where is version 0.8.0?", closely followed by "where is version > 1.0.0?" > > You basically tie the version cycle of the software to the version > cycle of the spec. > > Please, use apr versioning (or linux versioning or whatever three > digit versioning is considered these days) and start with a sane > (e.g. 1.0.0) version. > is there any conflict between calling something '1.0.0' and still being in incubation? Is this something we want to reflect in the versioning or not? Though I also very much agree that a 1.0.0 sounds a lot more 'ready to be used' then a 0.9.0 release would indicate. > >Happy with the version numbers ? > > No, not at all. I am -1 on the 0.8.1 naming shebang. Let's see if that > count for something here. it's an apache project, so that pretty much goes without saying, though it's good to note that that's a non-binding -1 vote (unless i'm reading the committers file incorrectly), but that's semantics
