web client: attempting to update a MARC record without required permissions fails without feedback to the user

[Kathy Lussier]

Tested on 2.12

If a user account does not have the UPDATE_MARC permission and attempts to perform an edit on the record, the update action will fail without any feedback to the user. In the XUL client, the user will receive an error message at the point that the user tries to save the record.

Ideally, the user will receive a message as soon as they click the MARC Edit tab to prevent them from making a long series of edits that ultimately fail. However, for feature parity with the xul client, at a minimum an error message should pop up when they click Save.

[Michele Morgan (mmorgan)]

Related: bug 1716765

[Bill Erickson (berick)]

I tracked the failure to report to the user back to bug #1808016. However, the bug could also be avoided by checking permissions in advance, when the page first loads.

[Bill Erickson (berick)]

Additionally the MARC editor should probably use the 'open-ils.cat' API's since they have additional smarts baked in for handling TCN collisions, etc. This would also resolve the silent-failure permission issue.

[Bill Erickson (berick)]

Marked bug #1716765 as a duplicate.

Plan is to migrate to using open-ils.cat APIs and to add notifications (toasts) on save/create/delete success or failure.

[Bill Erickson (berick)]

Fixes pushed:

https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/berick/lp1693580-marc-editor-api-notify

Note there are 2 commits. One fixes some Perl API bugs, the other applies the UI changes.

    1. Indicate in the interface when a bib/auth MARC record edit failed or
    succeeded.

    2. Handle permission failures on MARC edit by displaying the permission
    override dialog.

[Bill Ott (bott)]

After applying this, attempting to undelete a title fails. I'm unsure if there's any relation, but wondered if anyone else could replicate the issue, before I started chasing red herrings.

[Bill Ott (bott)]

I can verify that this had nothing to do with the undelete issue.

[Dan Pearl (dpearl)]

I have observed the error pop-up. I disagree with the point about feature parity with XUL; by all rights, any attempt to edit the MARC should come up with the error. It does not serve the user to make edits, just to have to discard them. Regardless, it does behave as intended, and I will sign off.

user/dpearl/lp1693580-marc-editor-api-notify-signoff

[Dan Pearl (dpearl)]

Retracting the sign-off for further testing (I omitted testing one commit)

[Dan Pearl (dpearl)]

Reinstating the signoff after retesting and proper inclusion of a second commit.

[Bill Erickson (berick)]

Thanks, Dan. Did a quick re-test and merged to 3.2 and above.

[Bill Erickson (berick)]

Setting 3.1 to WON'T FIX, since the patch does not cleanly back port to 3.1.