Stop passing exportMode, componentTable and ids by reference in export hook

[seamuslee001]

Overview

As per discussion on a recent export ticket it was discovered that its practically impossible to alter exportMode, ComponentTable and the ids array according to @MegaphoneJon #15266 (comment) and backed up by @scardinius #15266 (comment)

ping @MegaphoneJon @eileenmcnaughton @scardinius

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[MegaphoneJon]

I would add $exportMode to this list. I also neglected to add $ids to the list of arguments that would trigger a CRM_Core_Error::deprecatedFunctionWarning() if modified.

[seamuslee001]

done @MegaphoneJon

[seamuslee001]

Jenkins re test this please

[MegaphoneJon]

I reviewed this from both a code and r-run perspective, and have no issues. This change requires documentation, which I just submitted.

[totten]

Regarding this explanation:

As per recent Export discussions it seems we don't need to pass these 2 variables by reference

r-explain is generally pretty relaxed - if the substantive explanation is in a link to Gitlab or StackExchange or some wiki, that's OK. However, IMHO, the above text is too ambiguous. Put yourself in the shoes of an extension-author reading the release notes/change log and viewing this page. It would seem like we were removing functionality from the hook (e.g. can no longer alter the ID list) for no reason. It just raises a bunch of questions that the reader can't answer. To wit:

Why don't we need to pass that by reference? Is it because it never really worked? Maybe the old code was over-promising/ill-conceived? Or maybe it worked, but it became irrelevant because of some other change? Or maybe we realized that the functionality causes more harm than good? Or we found that there was no person (in MM) and no code (in universe) which actually relied on it? If I want to find out what the reason is... which discussion should I read? How do I find it? Search Google? Search Mattermost? Search Gitlab? Search Github PR? Should I even bother looking for the discussion - will it give me a clear explanation, or will it take 40 minutes to read the back-and-forth among different POVs?

I'm not saying that this particular issue needs a long explanation - but, generally, r-explain does require some explanation.

[seamuslee001]

@totten i have fleshed out the description a bit

[totten]

Cheers @seamuslee001 👍