Update old MySQL packages

[rzats]

Closes #1355.

Summary:

Bumps versions for the mysqlclient and SQLAlchemy libraries.

Removes the mysql-connector and pymysql libraries, replacing all their usage with the MySQLdb interface.

Prerequisites:

• Unless it is a documentation hotfix it should be merged against the dev branch
• Branch is up-to-date with the branch to be merged with, i.e. dev
• Build is successful
• Code is cleaned up and formatted

[melange396]

did you see the comments i added to #1355 yesterday?

[melange396]

Nice, this is pretty much a drop-in replacement except for the cursor name accessor. Does this cursor.rowcount still work with the new adapter? If the method exists, i would assume so, but you never know.

delphi-epidata/src/acquisition/covidcast/database.py

Lines 175 to 181 in 8a476fd

	modified_row_count = self._cursor.rowcount
	self._cursor.execute(fix_is_latest_issue_sql)
	if not suppress_jobs:
	self.run_dbjobs() # TODO: incorporate the logic of dbjobs() into this method [once calls to dbjobs() are no longer needed for migrations]
	if modified_row_count is None or modified_row_count == -1:
	# the SQL connector does not support returning number of rows affected (see PEP 249)

We talked about performance differences between the adapters -- i expect this to be approximately the same as with the previous one if not a teeny bit better. But just in case that doesnt hold, it would be good to try timing some stress-testing. From a simpler perspective, how does the running time of the integration tests compare to before?

If you have time, could you try to see if this adapter would support the reeeeeally long INSERT INTO...UPDATE statement that broke things before #1356? I have a strong feeling that the old "upsert" is better for data file fragmentation than the new REPLACE INTO.

[rzats]

@melange396 the rowcount property is available in the new cursor API, so no issues there :)

As for a quick performance check, let me try:

/run performance test

[github-actions]

✅ Performance tests complete! Result summary:

• Total requests: 1000
• Total failures: 12
• Min response time: 7.69 ms
• Max response time: 128314.30 ms
• Average response time: 1334.67 ms
• Median response time: 380.00 ms
• Requests per second: 3.17

Click here to view full results: https://github.com/cmu-delphi/delphi-epidata/actions/runs/7572075216.

[melange396]

As for a quick performance check, let me try:

i dont think that demonstrates very much, since its not doing acquisition or even going through the integration tests

[sonarqubecloud]

Quality Gate passed

Kudos, no new issues were introduced!

0 New issues
0 Security Hotspots
No data about Coverage
8.1% Duplication on New Code

See analysis details on SonarCloud

[melange396]

Do you have any updates on performance comparison and/or compatibility with the old long INSERT INTO...UPDATE statement?