Avoid bad database state after failed migration

[rfwatson]

Ensure g.repo is not nil after a failed migration, this allows the cleanup here to be triggered as expected.

Fixes #7039

[rfwatson]

Is there any docs for setting up integration tests locally? I'll add a test case if I can get them running

[zeripath]

make test-sqlite will run the integration tests on sqlite which is probably all you need to run unless you're very keen.

I think CONTRIBUTING.md has some information about how to run drone to perform a pull-request check.

drone exec --local --build-event pull_request

[rfwatson]

Thanks.

Out of interest, is there a reason we don’t rollback a database transaction here, instead of cleaning up manually?

[zeripath]

It's not in a transaction unfortunately. Keeping it in a transaction would cause the entire database to be locked whilst you're migrating potentially large repositories.

[codecov-io]

Codecov Report

Merging #7040 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #7040      +/-   ##
==========================================
- Coverage   41.45%   41.44%   -0.01%     
==========================================
  Files         442      442              
  Lines       59597    59597              
==========================================
- Hits        24703    24701       -2     
- Misses      31665    31667       +2     
  Partials     3229     3229

Impacted Files	Coverage Δ
modules/migrations/gitea.go	8.3% <100%> (ø)	⬆️
models/repo_list.go	72.08% <0%> (-1.02%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 355ab0c...e7240e9. Read the comment docs.

[lunny]

@rfwatson good catch.