Apply the same conditional copying of cached sources as cached tools

[sbesson]

Platform builds have started failing recently since there is no more sources
in the cache. This commit copies the logic used for the cache tools and should
prevent failures if no source exists.

The expectation is that this will fix https://ci.openmicroscopy.org/view/Failing/job/OME-FILES-CPP-DEV-merge/

[rleigh-codelibre]

Why are there no sources in the cache? It should be downloading at least one source file each time it runs (gtest), or using that file from the cache. What changed to make this break?

[sbesson]

@rleigh-codelibre exactly, the underlying issue seems to be that gtest is no longer detected/built and cached. I will go through the logs and see where it stopped working.

Update: from my debugging, the gtest detection stopped working in https://ci.openmicroscopy.org/view/Failing/job/OME-FILES-CPP-DEV-merge/471/. This might be related to the changes brought in #126

[sbesson]

Looking more carefully at the logs, is this because #126 introduced a breaking change in the project name from ome-superbuild to ome-cmake-superbuild? Thus the ome-superbuild_xxx options should now be renamed as ome-cmake-superbuild_xxx esp. ome-superbuild_BUILD_gtest used in the Jenkins script?

[rleigh-codelibre]

Ahhh, yes that would account for it. I've opened #129 to correct this.

[sbesson]

Thanks. Closing this in favor of #129, I'll re-run the builds

[rleigh-codelibre]

@sbesson I think having this logic is helpful to prevent any future breakage with changes to the caching behaviour, so feel free to leave it open.

[sbesson]

@rleigh-codelibre https://ci.openmicroscopy.org/view/Failing/job/OME-FILES-CPP-DEV-merge/477/ is green with #129 included. I could get this merged to make the script more robust. On the other hand, the breaking job failure was indicative of an underlying issue with the project renaming. Happy to go either way.

[rleigh-codelibre]

I think merging both would be the best approach.