I'm trying to get a custom-install handler packaged as part of a
headless RCP build. I'm encountering a problem during the packaging
phase where the P2 system executes the install action of the custom
install handler. This, in turn, causes the build to fail as it attempts
to execute a file that does not exist at this stage.
The reasoning behind this setup is that we have an update-site available
where our users can install or upgrade a package of binary files. The
stand-alone installer can be either delivered directly, or downloaded
and run through the eclipse update system. Our product runs on Mac,
Linux and Windows so there is a feature for each of these platforms
which contains a platform specific version of the installer binary. I
have a grouping feature which holds all three platform features as well
as the custom install handler plug-in and puts the OS dependencies on
them so that the correct installer is downloaded and executed during an
upgrade from any of the platforms. This process works fine when done
through the Install new software, but when doing a headless build we do
not want to run these installer binaries, but simply include the feature
so the user can later just check for updates to get the latest versions.
Is there any-way to detect that the install handler action is being
called from a headless build and to simply return out of the install
action code? Is there a better way to do what I'm attempting?