I've been struggling with the following workflow:
I have two codebases (separate github repos). The first produces a package that is consumed in the second codebase. On a day to day basis, the first codebase changes very rarely, so we publish the first codebase to a package which gets downloaded when building
the second. Nothing interesting here.
But what if I want to test that a change to the first codebase won't break the second? This workflow seems really difficult to achieve. What I'd really like is to do something like:
- Install packages based on the nuget config, but for package X, look at a local source instead of the regular feed.
You could theoretically do this by replacing the source in the nuget config with the local source, but what if that feed also had other packages in it that are needed? Then all the dependencies may not be satisfied. You can add the new source without disabling
the old. but then you don't know whether you're getting the official version or the testing version.
Simply reinstalling or restoring over the previous version with the new source seems like it would be a good candidate, but there is no "force reinstall". I'm only left with the option of deleting the folder containing what I want to replace, then
installing. This is clunky and hacky.
It seems like this is a missing feature. It's really difficult to test packages before deploying them. Even a test feed falls short here, since your test feed must contain all the original dependencies you wanted, plus the delta to your new package.
Powershell and VS are out too, since I need this to work on mono.