And when vendors force our dismembered hand into breaking changes we should probably update major version.
On Thursday, October 4, 2012, Brian LeRoux wrote: > oh yes. recent changes that burn the build team come to mind too. > > the policy we're gunning for is we never break anything, we DEPRECATE > and warn for 6 months. > > the practice might vary of course when we're dealing w/ vendor missteps. > > On Thu, Oct 4, 2012 at 7:48 PM, Filip Maj <f...@adobe.com> wrote: > > Indeed, all of us in this project need to be more careful about that, Bri > > remember your "cut off your arms and legs and left you by an unforgiving > > flow of magma" blog post on phonegap.com ? > > > > On 10/4/12 10:36 AM, "Mike Reinstein" <reinstein.m...@gmail.com> wrote: > > > >>> I don't think we have the luxury of knowing when something breaks > >> > >>Granted, and that's not something we can really fix. *However, we can > >>identify when our API changes in breaking ways*. > >> > >>-Mike > >> > >>On Thu, Oct 4, 2012 at 6:37 AM, Brian LeRoux <b...@brian.io> wrote: > >> > >>> I personally don't think semver really did fix anything in ruby-land > >>> (but thats my opinion). Ruby has a crummy package system.The only one > >>> worse is Pythons. > >>> > >>> Anyhow, I added a little bit about our releases in the wiki [1] and a > >>> much longer post to the phonegap blog [2] to help folks better > >>> understand the rational. To echo Fil, I don't think we have the luxury > >>> of knowing when something breaks given the cat and mouse nature of the > >>> project relationship to mobile operating system vendors. > >>> > >>> [1] http://wiki.apache.org/cordova/CuttingReleases > >>> [2] > >>> > >>> > http://phonegap.com/2012/04/12/rolling-releases-how-apache-cordova-become > >>>s-phonegap-and-why/ > >>> > >>> On Wed, Oct 3, 2012 at 11:44 PM, Mike Reinstein > >>> <reinstein.m...@gmail.com> wrote: > >>> > I certainly don't meant to rehash something that has been discussed > >>> > ad-nauseam. Nor am I advocating we change how often we release. I > >>>think > >>> the > >>> > key distinction is picking a version number that indicates breaking > >>> change, > >>> > compatible changes/new features, vs patches. Semantic versioning > >>> provides a > >>> > clean way to do specify this. In npm and ruby land, this has largely > >>> fixed > >>> > dependency hell, and has led to more reliable code re-use. > >>> > > >>> > Just a thought. > >>> > > >>> > http://semver.org > >>> > > >>> > > >>> > On Wed, Oct 3, 2012 at 5:37 PM, Filip Maj <f...@adobe.com> wrote: > >>> > > >>> >> This discussion again :) > >>> >> > >>> >> http://apache.markmail.org/thread/l2et3r5v35efprgd > >>> >> > >>> >> > >>> >> With a point release coming out every month or so that limits us to > >>> being > >>> >> able to "break" things every 10 months or so. With changing SDKs > (see > >>> iOS > >>> >> 4.2, 5, and 6) sometimes we need to break things, like, asap. > >>> >> > >>> >> Other times we break things because we are assholes (from our users' > >>> point > >>> >> of view, at least :P ) > >>> >> > >>> >> On 10/3/12 2:21 PM, "Mike Reinstein" <reinstein.m...@gmail.com> > >>>wrote: > >>> >> > >>> >> >I'm wondering if anyone else has given thought towards adopting > >>> semantic > >>> >> >versioning for our releases. In terms of making plugin development > >>>and > >>> >> >version adoption less painful, this might be a good move. Thoughts? > >>> >> > > >>> >> >-Mike > >>> >> > >>>