Closed stelfrich closed 5 years ago
To know whether a warning is sufficient—or whether a property configuring behavior is necessary—we need to know the use cases to which the mojo will be put.
For end users, testing stuff in a Fiji installation, I think a warning is sufficient.
For automated systems deploying updates to update sites (see e.g. how SciView does it), it may make more sense for the build to fail when anything non-SemVer-compatible is detected.
For end users, testing stuff in a Fiji installation, I think a warning is sufficient.
I guess by end users you mean developers in this particular case, @ctrueden?
For automated systems deploying updates to update sites (see e.g. how SciView does it), it may make more sense for the build to fail when anything non-SemVer-compatible is detected.
This could be handled by an imagej.failOnIncompatibleVersion
flag that is set to false
per default. I think warnings should be logged in both cases and the plugin should only fail if imagej.failOnIncompatibleVersion==true
.
@stelfrich Could you migrate this to scijava/scijava-maven-plugin if it is still applicable?
Closed in favor of https://github.com/scijava/scijava-maven-plugin/issues/22.
Would it make sense to at least log a warning when overriding an older version of an artifact that is incompatible (according to SemVer) with a newer one? Should this be an extra property?
Does anyone have an opinion on this, @imagejan, @ctrueden?