I''m submitting a patch that fixes the 2 test failures below.
The first was the result of incomplete handling of camelCase column
names in Oracle. Since camelCase column names are kind of a no-no in
Oracle anyway, I was tempted to "fix" the failure by changing the
tests
to just skip this test when testing Oracle. But it turned out the fix
wasn''t too bad, and taught me something about Rails.
The second was the result of how the OCIAdapter makes assumptions about
whether a column is a Date or a Time. Since that isn''t what this
specific test is about (it was checking the protection of attributes in
mass assignment), I cheated -- I fixed it by changing the actual time
used so that it included a non-zero hour, which makes it work.
I may put finding a better solution to the Date/Time problem on my list
as well.
Michael Schoen wrote:> In working on my patches to the oci adapter, I find that the trunk
> currently fails the base_test suite. I''ll take a look, but if
somebody
> with more experience knows how to fix these...
>
> (or am I just doing something wrong?)
>
> $ ruby -I connections/native_oci/ base_test.rb
> Using OCI Oracle
> Loaded suite base_test
> Started
>
...F.....................................................F..............................
>
> Finished in 19.711216 seconds.
>
> 1) Failure:
> test_attributes_hash(BasicsTest) [base_test.rb:117]:
> <{"extendedWarranty"=>1, "id"=>1,
"developer"=>1}> expected but was
> <{"id"=>1, "extendedwarranty"=>1,
"developer"=>1}>.
>
> 2) Failure:
> test_multiparameter_mass_assignment_protector(BasicsTest)
> [base_test.rb:578]:
> <Sat Jan 01 00:00:00 PST 2000> expected but was
> <#<Date: 4903089/2,0,2299161>>.
>
> 88 tests, 197 assertions, 2 failures, 0 errors
>
>