I'm undecided about this patch series ☹ - feedback needed. It's goal is to take advantage of the fact that htools can use “offline” data files and thus doesn't need access to a live cluster. That it does well.
The downside is that, not having a nice shell-level testing framework, this is all done in a hackish way, and if the tests fail you can't do anything except use "bash -x" and hope for the best. The other thing I'm not 100% is whether the tests we do are correct/fair enough to warrant inclusion in the coverage data; with this patch series, coverage for the htools codebase reaches ~60%, and it seems too easy to reach this just via 20 command invocations. Iustin Pop (6): Add a helper for htools CLI tests Add a simple binary tester for htools Enable htools offline tests as Haskell tests Build a coverage-enabled version of the tools Enhance the offline tests Add more offline tests .gitignore | 2 + Makefile.am | 27 ++++++++--- autotools/run-in-tempdir | 8 ++- htools/cli-tests-defs.sh | 44 ++++++++++++++++ htools/htools-hpc.hs | 1 + htools/live-test.sh | 82 ++++++++++++++++--------------- htools/offline-test.sh | 122 ++++++++++++++++++++++++++++++++++++++++++++++ 7 files changed, 236 insertions(+), 50 deletions(-) create mode 100644 htools/cli-tests-defs.sh create mode 120000 htools/htools-hpc.hs create mode 100755 htools/offline-test.sh -- 1.7.9.1
