[Nitro] Nitro/OG spec questions

Mark Van De Vyver mvyver at gmail.com
Fri Sep 7 04:35:14 EDT 2007


Hi,
As you might have seen from the previous email, I've been wrestling
with specs for the DBI Adapter - I am restricting my self to adapter
behavior, but still run into problems that whoever has or will spec
the Manager, etc classes, is likely to encounter.

Specifically, ensuring that you 'start' each spec in exactly the same
state, i.e. the spec should be atomic/uncontaminated by pervious
results.
This seems to be the 'spec-best-practice' - correct?  I can believe
that because I can imagine a debug head ache if you start getting
false negatives or the worst - false positives.

Initially I thought to use `describe do ... end` for every spec where
the options passed in are different.  Even this does not help since
the changed attributes (class variables) persist between 'describe'
specs.

I'm not familiar with the spec infrastructure - I assume starting a
spec in a new file ensure your starting 'from-scratch'?  Have I missed
something obvious?

Choices:
stick with single spec file per source, and just wrestle (and cross fingers).
split the sources and maintain on spec file per source file (may/ may
not fully solve the issue?).
adopt some convention for multiple spec files per source (pointless if
this doesn't over come the above problem).

Comments/thoughts/tips?

Mark


More information about the Nitro-general mailing list