Code review comment for lp:~epics-core/epics-gateway/dbe_props

RHEL 6.6 ships with Python 2.6.6 so that made a major difference, would you like me to commit those changes?

I still have test failures under 3.14, but not 3.15 — could you maybe check for the Base version and disable these tests under 3.14? With epicsUnitTest I would mark these tests as SKIP:

FAIL: DBE_PROPERTY monitor on an ai - value changes generate no events; property changes generate events.
FAIL: Get PV (value) through GW - change HIGH directly - get the DBR_CTRL of the PV through GW
FAIL: Monitor PV (value events) through GW - change HIGH directly - get the DBR_CTRL of the PV through GW

Hmm, when I run nosetests manually in the pyTestsApp directory, pyepics can't find libca.so at all. I think make is setting the EPICS_BASE environment variable which is sufficient for pyepics to work. You should probably mention the need to set EPICS_BASE in the testTop/README.

Makefile issues:

If I use 'make runtest' it re-runs the tests (on the host, using the host binaries) for every cross-target architecture I have Base configured to build. The Makefile should probably use "ifeq ($(T_A),$(EPICS_HOST_ARCH))" instead of "ifdef T_A".

Please use $(INSTALL) or even $(CP) instead of install, which is not portable. The final argument to $(INSTALL) must be the directory name to install into though, not the target filename.

Your build is doing installations of the .py files at 'make runtests' time, not during the initial 'make'. That works, but you might want to put them in TESTSCRIPTS_HOST instead of your own PYTESTS variable and then they will be installed automatically at 'make' time (you do still need the %.py: pattern rule though). That seems to partially fix the issue that 'make clean' doesn't work from inside the O.<host> directory.

I don't understand your comment in the Makefile about making stuff "go through common dir".