[danielrazavi]

Follow-up to #481: accept different valid SQL answers without accepting incorrect results.

Fixes

Feedback from	Issue	Fix / commit
@khsergvl	Different column names or order could fail	Match values, not names; still check which values belong together and how often rows appear. 18c75f8
@niyazmnazari	Q4 required unnecessary purchase details	Vendor/time columns are optional; clearer feedback. e126332
@niyazmnazari	Ordinary -- query comments caused errors	Only complete answer markers are treated as markers. 1e7d3d4
@niyazmnazari	Q9 temporary tables failed and blocked Q10-12	Temporary tables now work throughout Q9-12. 97ea818
@niyazmnazari	Q4 rejected date-only running counts	Accept counts grouping purchases by date or identical timestamp. a3cd528
@niyazmnazari	Q3 required displaying the rank	Rank can stay inside the query. 387711b
@niyazmnazari	Q8 revenue needed a specific name/position	Revenue can have any alias and appear before extra columns. 15be64b

Both A2 versions (DC Cohort and Microcredential Cohort) now remind students to track names used in earlier queries, rather than impose output-name restrictions.

Cleaner feedback: correct answers' SQL/output is collapsed; wrong answers and errors stay expanded. Status/messages always remain visible. bb2e41b

What I Tested

• Ran 31 automated tests with the private instructor solutions; all passed.
• Submitted correct answers with every Q1-8 output column renamed and reordered; all 12 questions passed.
• Deliberately mixed up Q2 customer/date pairings without changing the individual values; Q2 correctly failed.
• Submitted all-wrong answers; Q1-9 failed and Q10-12 were blocked because Q9 was wrong.
• Confirmed new commits automatically rerun grading, correct answers collapse their SQL/output, and wrong answers stay expanded.

No sample answers or instructor solutions are added to the course repo.

[github-actions]

Thanks for your contribution! 🎉

Please remember to tag or request a review from the DSI team. Give us up to 72 hours to review your pull request. We appreciate your patience and efforts.

[danielrazavi]

@khsergvl @niyazmnazari thanks again for the feedback! I've addressed the points you raised and tested the changes in my fork, including renamed/reordered columns and a deliberately wrong answer to check it still gets rejected. The description links each fix and the test results. Could you take a look and review when you have a chance?

[khsergvl]

@danielrazavi there is a need to make this part abstract:
2

And here:

https://github.com/UofT-DSI/sql/blob/main/.github/workflows/sql_assignment_runner.yml#L93

[danielrazavi]

@khsergvl we kept those parts specific to A2 because that’s the only assignment the grader supports right now. Once we add support for other assignments, we can make them configurable rather than hardcoded.

[khsergvl]

Looks good! Thank you so much!

[khsergvl]

@khsergvl we kept those parts specific to A2 because that’s the only assignment the grader supports right now. Once we add support for other assignments, we can make them configurable rather than hardcoded.

Just FYI khsergvl#13

[danielrazavi]

Now that the A1 restoration is part of this branch and PR, I'm moving this back to draft while I test both Assignment 1 and Assignment 2 in my fork. I'll mark it ready for review again once testing is finished.

[khsergvl]

Great work!