Fixing remaining issues of odbc metadata queries #94
Conversation
All implemented functions have a SQL Fallback in case pyodbc is not used. get_foreign_keys currently has the problem that it seems that at the moment the metadata function SQLForeignKeys does not work.
…ns and sql fallback
first deadlock tests
… environment startup
…sn't find the schemas
…is doesn't return a schema, if this schema doesn't contain tables
Furthermore, added tests for has_table and test against multiple engines with different default databases.
…ause a deadlock anymore
…nsformation from return statements into variables
tkilias
changed the title
|
@BY-jk So, I now got all tests green. There were still some edge cases left which made trouble, but now it should be fine. The previous pull request #90 got closed, probably by an accidental push to master, which I reverted within a minute. If you are ok with this state, I would merge it to master, and then we can discuss, how we get this to PyPi. |
|
@tkilias excellent work. Please go ahead and merge to master.
@nndo1991 is working on getting the GitHub Action workflow ready to release. |
|
Hi @tkilias is this the final version of your odbc changes or still planning to do more? I'm almost done on the Github Actions setup in my forked repo and will work on releasing 2.1.0 in the next couple of days. |
|
Hi @nndo1991, that is for the near future the final version. It avoids for pyodbc the deadlocks for all metadata functions. It shows for the tests the same behavior as the previous implementation and I specifically added tests to compare their results and to check the deadlocks behavior. The only difference happens in case of concurrent DDL operations, but this is normal for snapshot behavior. We are also thinking about a general solution within the database core, but this will take a while. So, from my side you can publish the current state. Thx. |
This is continuation of pull request #90 and fixes the remaining test failures.
Closes #89