Make MS-SQL / pyodbc unit tests pass

Issue #481 resolved
paj created an issue

I am planning to make the unit tests pass. With the first patch attached here, the failures are down to:

Engine: FAILED (failures=2) SQL: FAILED (failures=4, errors=4) ORM: FAILED (failures=4, errors=7)

With patch #509 and the second batch of fixes, the total failures are down to: FAILED (failures=5, errors=4)

Comments (12)

  1. Former user Account Deleted

    (original author: ram) I committted most all of this except for changes to sort order that appeared to be related to eagerload fetching, after noticing Mike posting a notice about changes in the ORM to handle this.

    Changeset 2412

  2. Mike Bayer repo owner

    I think this is OK for now, but lets not go too crazy with the somexistingtest_mssql thing in the unit tests. particularly in the ORM where we have so many tests already and its already a real mess in need of a good cleaning.

    it might be better in some cases to just take some of the existing tests and split them up into more granular tests (like the one in mapper.py with "limit" and "offset", split up into one that does just "limit"; then put unsupported(mysql) on the one that has "offset").

    i still wonder if we should start building db-specific test modules/directories too, so that db-specific stuff has a home of its own.

  3. Former user Account Deleted
    • assigned issue to

    (original author: ram) Committed (mostly, see notes below), in rev 2430, thanks.

    Note I did not commit the changes to ORDER BY in ansisql - I want Mike to have a look at those as they are likely to have a farily large effect. Everything else went in. Leaving the ticket open for now.

  4. paj reporter
    • assigned issue to

    I'll take discussion of ORDER BY to ticket #509

    Here's a patch that solves one more issue. For pyodbc at least, nearly all the tests pass now. There's just one nagging issue with Unicode, that will be a ticket in its own right.

  5. Log in to comment