Just created an issue for swapping to cordovaDependency from engine. https://issues.apache.org/jira/browse/CB-13123
I agree that we should switch the behavior to fail. It would need to be a major version of cordova. Chatted with Shaz & Simon and decided it would make sense to add a flag to current cordova that allows plugin add to fail if the dependency fails. So cordova@7 the default behavior will be skip the plugin or dep that doesn't satisfy the requirement and fail will be behind a flag. Once we hit cordova@8, we can swap the behavior so fail is default and skip is behind the flag. I will update https://issues.apache.org/jira/browse/CB-12122 with this proposal. Let us know what y'all think. On Wed, Jul 26, 2017 at 12:50 PM, Filip Maj <maj....@gmail.com> wrote: > OMG late to this thread but yes, this should fail and freakout. My > assumption is the same as the person who reported CB-12122 is: any > dependent plugins that fail constraint checking would bubble the > failure up the dependency chain. Unfortunately, the behaviour is not > _super_ explicit in the docs [1] - we should fix that. Currently it is > "The CLI aborts with a non-zero code for any plugin whose target > project does not meet the engine's constraints." To me that implies > fail+freakout, but we could elaborate on the wording here a bit more. > > This issue is closely entangled with how to integrate certain plugins > into platform code, esp. if the plugin is a common dependency. In > particular looking at cordova-plugin-compat [2]. The decision in this > thread would determine whether future versions of core plugins that > currently depend on compat (camera, geo, and others) would need to > depend on compat forever, or could remove the dependency altogether. > If we keep current behaviour, future versions of e.g. camera need to > keep a <dependency> on compat. If we change the behaviour to fail and > freak out, we can remove the dependency after integrating compat into > cordova-android. Strict engine constraints would need to land in > plugins moving forward in either case. > > I would also like to push for aligning cordovaDependency and <engine> > behaviour ASAP. As per our docs today [3], we already say "[the > cordovaDependencies] feature is intended to eventually replace the > engines element in plugin.xml." Do we have an issue filed to do that? > > [1] https://cordova.apache.org/docs/en/latest/plugin_ref/ > spec.html#engines-and-engine > [2] https://issues.apache.org/jira/browse/CB-12730 > [3] https://cordova.apache.org/docs/en/latest/guide/hybrid/ > plugins/index.html#specifying-cordova-dependencies > > On Fri, Jun 2, 2017 at 12:15 PM, Kerri Shotts <kerrisho...@gmail.com> > wrote: > > +1 Fail (I think principle of least surprise applies here; most devs > would expect an incompatible plugin to fail the build.) > > > > ~ Kerri > > > >> On Jun 2, 2017, at 12:15, Shazron <shaz...@gmail.com> wrote: > >> > >> Consensus on this long stewing issue of 7 months? > >> > >> If you deleted the thread: > >> 1. Issue https://issues.apache.org/jira/browse/CB-12122 > >> 2. Thread https://s.apache.org/ofqR > >> > >> +1 on engine check failing the build. > >> Questions that need answered: Does cordova-fetch on cordova@7 do this > >> already, like Simon said? > >> > >> > >> On Mon, Nov 7, 2016 at 6:34 PM, Simon MacDonald < > simon.macdon...@gmail.com> > >> wrote: > >> > >>> I think an engine check should fail the build. If there plugin version > >>> they are trying to use will not work with the version of cordova or > >>> platform they are currently using in their project it should not > >>> build. Folks can screw up their semver in config.xml and end up > >>> pulling in a plugin version that doesn't work with the CLI they are > >>> currently using so stuff will just be horribly broken without them > >>> knowing. > >>> > >>> Also, I thought the new cordova-fetch was supposed to prevent this > >>> sort of thing? > >>> Simon Mac Donald > >>> http://simonmacdonald.com > >>> > >>> > >>> On Mon, Nov 7, 2016 at 8:24 PM, Shazron <shaz...@gmail.com> wrote: > >>>> See https://issues.apache.org/jira/browse/CB-12122 > >>>> > >>>> Right now if a plugin fails an <engines> check, it just warns. Should > we > >>>> make engine checks fail the build? > >>> > >>> --------------------------------------------------------------------- > >>> To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org > >>> For additional commands, e-mail: dev-h...@cordova.apache.org > >>> > >>> > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org > For additional commands, e-mail: dev-h...@cordova.apache.org > >
