Re: RFC: Testing datasets for non-sideeffecting test-specific data.

http://github.com/UnwashedMeme/clsql/commits/dataset-refactoring  for
browsing and reviewing

To merge the changes, should be a fast-forward merge:
git pull git://github.com/UnwashedMeme/clsql.git dataset-refactoring

Test results
* Mysql: mostly passes, the remaining failures I believe to all be
actual problems in the implementation of normalized classes.
* ODBC: postgres backend passes, sqlserver 2005 fails during the setup
of trying to truncate the database.  No normalization tests
* postgres-socket: everything passes, no normalization tests
* sqlite: many failures, fewer than before though.

I mostly just used the dataset feature to encapsulate the existing
view-class based test code. This is partly based on the difficulty in
creating any really cross-platform sql statements (a lot of the point of
all the backend code in clsql), partly just I didn't want to spend the
time rewriting everything ;-).

Let me know if anything looks awry or causes problems.

On 1/12/2010 8:11 PM, Kevin Rosenberg wrote:
> On Jan 12, 2010, at 1:11 PM, Nathan Bird wrote:
>   
>> Please let me know what you think. Is this something that could be
>> included in the main branch?
>>     
> Excellent idea and would love to.
>
> Kevin
>