Previously, when weighing the merging of wskdeploy into CLI, it seemed that the CLI was a good "base" for CRUD operations (commands) against the OW primitives... and that indeed wskdeploy was a great "plug-in" bringing in higher order concepts (some of which are not official OW primitives).
In addition, and as a result of the added complexity, the wskdeploy codebase is quite large and fear that it would both increase the complexity of maintaining the CLI, as well as reduce experimentation (e.g., perhaps towards leveraging other deployment targets like knative and bringing in source-to-image pipelines like Jib/Tekton). Still fall on the side of keeping the CLI simple and CRUD based around our existing primitives, but deserves continued discussion over time. -Matt > > From: Rodric Rabbah <rod...@gmail.com> > To: dev@openwhisk.apache.org > Date: 07/09/2019 03:13 AM > Subject: [EXTERNAL] Re: rationale for wskdeploy Docker image > > > > I think we don’t need it but defer to those who worked on it. > > I’d like to also ask at the risk of broadening the original question: > > - should we fold the repository into the cli? > - or is there a good reason to keep it as a separate deploy tool > independent of the cli? > > -r > > > On Jul 3, 2019, at 4:23 PM, David P Grove <gro...@us.ibm.com> wrote: > > > > > > Do we really need to be publishing a wskdeploy image to dockerhub? > > > > I'm not understanding why this image needs to be public (it appears to > > perhaps be an image for building wskdeploy?). > > > > thanks, > > > > --dave > > > > > > > >
