search

3246251196 / adtools_testing

1 stars 1 forks source link

Detection of SO objects should be more robust #9

Closed 3246251196 closed 1 year ago

3246251196 commented 1 year ago

At the moment, due to the way that the cross compiler installs libs, there is a hack: https://github.com/3246251196/adtools_testing/blob/80f2baabbcd7d2d6e317ae7fe9e28707546ead43/tests/common.mak#L40

This is needed to pull back the correct SO in for either clib2 or newlib. But, this is not really robust enough.

For one, the user may have a folder named "clib2" before the build directory. As in: /home/user/projects/clib2/adtools_testing... It is quite unlikely, but possible.

Also, but still unlikely, the user may have installed certain packages that place libs into the SDK. In that case, the framework should also pull those package SOs into the test: the correct ones from either CLIB2 or NEWLIB!

This ticket is to get around to improving figuring out precisely which SO to pull in to the framework.

3246251196 commented 1 year ago

Closed as mostly a duplicate of: #4

There is only so much that the script can do. Any Shared Objects created by the test are copied into the variant LHA. Shared Objects that created by the test/makefile and are not compiler/c-library are not copied and they will have to be manually added - hence the creation of EXTRA_FILES in the test framework.