Do we currently just test master plugins with master platforms? It would be nice to test master plugins against released platforms and master platforms with released plugins.
cordova-plugin-whitelist problem got solved in a unreleased fix to cordova-lib. It would have made sense to hold off on releasing this plugin until we did a tools release. I reverted the latest tag when this issue came up on the list. I wish I had caught the cordova-plugin-file-transfer bug. It was showing those tests failing over a few builds which made me think the fails were expected. Maybe we should have a expected fails per platform/plugin chart of some sort to quickly compare against? -Steve On Mon, Jul 6, 2015 at 10:16 AM, Nikhil Khandelwal <[email protected]> wrote: > https://issues.apache.org/jira/browse/CB-9248 highlights a testing gap > with our release process: > > * cordova-plugin-whitelist should not have been released with > having iOS platform >= 4.0 without iOS platform release. While reverting > the 'latest' tag was a good move to get this fixed, it would have been nice > to catch it earlier. > > * cordova-plugin-file-transfer should not have been shipped with a > build failure on WP8 because it references unreleased types in WP8 platform > in master. > While we are using buildbot to run and the build the plugin against master > version of the platform, most users are using it with the released version > of the platform. At the very least, this scenario needs to be tested at the > time of release, if not in a CI scenario with every commit. > > Would love to hear your thoughts on how we can catch these issues before > release? One of the ideas is to use buildbot to schedule a 'release' build > that would run the tests we have with the correct versions of the code that > we want - release tag for plugins against released versions of the platform. > > Thanks, > Nikhil > > >
