Viewing previous version does not show previous version accurately

[seabelis]

Problem

Sometimes I need to view the edit history of a record to see what changes were made and when. I count on the preview of the previous version to show me accurate information. Instead, the preview of a previous version shows changes that had not yet been made making it appear as though that data existed prior to the edit I'm reviewing.

Evidence / Screenshot

Diff view

Preview of v. 1

Actual v. 1

Relevant URL(s)

https://openlibrary.org/books/OL49822191M/Poesie?m=history

Reproducing the bug

1. Go to ...a record that has been changed incorrectly and does not have an accurate preview of previous versions.
2. Do ... see the inaccuracies as noted in the above example.

• Expected behavior: The version preview should display the previous version accurately.
• Actual behavior: It does not.

Context

• Browser (Chrome, Safari, Firefox, etc):
• OS (Windows, Mac, etc):
• Logged in (Y/N): Y
• Environment (prod, dev, local): prod

Notes from this Issue's Lead

Proposal & constraints

Related files

Stakeholders

@scottbarnes

[mekarpeles]

Can you help us understand a bit more clearly the issue with previewing old pages in this issue?

Are you clicking the view button? Or the revision 1 link? or the revision 2 link? And what data are you seeing that you shouldn't be? Or what data aren't you seeing that you should be? What do you expect v. what is happening?

[seabelis]

When viewing the preview of rev 1. I see changes that were made after that version. If I actually revert the record to v.1 I can see actual v.1. You can see in the screenshots the different publisher. That publisher was not added until the edits that resulted in v.2. Does that help?

[mekarpeles]

Alright, so the problem is, the preview feature may be off by one and showing that version plus the changes made on top of it for the next revision. I think this needs investigation, though I think I better understand what seems to be happening. Given this, I think we'll stage in 2024-06 milestone as P2

[mekarpeles]

@seabelis After having a team call, we came to the conclusion that specifically the editions table is what seems misleading here, because the contents of the editions table are coming directly from infogami and are no responsive to the edit history / version control.

In other words, when looking at the diff, the information in the editions table should be ignored and the edition details is the thing we should rely on.

We could make it so if there is a ?v= parameter in the url, that the version of the edition that has been used to load the book page gets passed into the editions table (that we load from solr or whatever) so that after the overriding occurs, the right data is shown in the editions table as the rest of the page.

For now though, let's deprioritize as P3, if someone from the community wants to shim this in to our existing setup, feel free, the simple answer is that when we explore revisions of books, the info in the edition table should be ignored. However, anyone who does want to contribute should coordinate with @jimchamp who is driving #7451