Improve logging in project deletes

[saipavan10-git]

This pull request includes significant changes to the delete_project function and its associated tests. The changes enhance logging, and update the test cases to reflect the new logging structure. Addresses #169

I tested the new delete function and now the project delete is visible on REDCap UI.

Enhancements to delete_project function:

• Enhanced logging by adding detailed log parameters such as ts, user, ip, page, event, object_type, sql_log, pk, event_id, data_values, description, legacy, and change_reason.

[pbchase]

See comments

[pbchase]

Test this with an invalid username to see how REDCap's UI will behave.

Then, revise the code to default it first to, the value in sys.getenv("API_USER"), then--if invalid usernames are allowed--the script name.

[pbchase]

Use getip::getip("local") to get the local IP address. Make getip a required package for redcapcustodian.

[pbchase]

@saipavan10-git, note that the checks are failing. If you drill in on that error you will see the error is pretty obvious: redcapcustodian now needs getip and the container used by the github action does not have that. I've fixed these before, but I should hand this one to you. I'll show you path, so you can fix it.

Could you document this procedure in addition to fixing it? The procedure doesn't seem clear unless you've already set up GitHub actions to perform automated testing. Any thoughts on where to put this procedure? Could we add a vignette for developers?

You can see the configuration for the action that failed at run-tests.yaml. It uses an image at ghcr.io/ctsit/rstudio-ci. The Dockerfile that defines that is at Dockerfile. Historically, we have added required packages in this Dockerfile. That should work quickly and easily, but I think you can install packages in run-tests.yaml that might be a better way to make the package self-documenting, but that might also make tests slower to run.

[saipavan10-git]

I updated this dockerfile thinking this will fix the issue.

Adding a guide vignette for developers is a great idea. I can create one to document the fix for this issue. If we add new dependencies more frequently, we could add a step to setup dependencies in run-tests.yaml like in pkgdown.yaml. This will reduce the need to update the rstudio-Dockerfile. Also, setup r dependencies has an option to cache the packages that might help with the speed of the gh action.

[pbchase]

I updated this dockerfile thinking this will fix the issue.

That's necessary, but not sufficient to make CI-tests work

Adding a guide vignette for developers is a great idea. I can create one to document the fix for this issue. If we add new dependencies more frequently, we could add a step to setup dependencies in run-tests.yaml like in pkgdown.yaml. This will reduce the need to update the rstudio-Dockerfile.

Right! I knew we had done this before, but I couldn't locate it.

Also, setup r dependencies has an option to cache the packages that might help with the speed of the gh action.

That minimizes the downside of the more local management of dependencies. I like it :-)