Roman numerals not sorted properly in ToC

[jordanallen-dev]

https://pep-web.org/browse/NLP/volumes/79?preview=NLP.079.0001A

The client expects the server to provide Roman numerals in order and will render them as-is with the exception that they will be put to the top of the ToC

In the above example, the ToC is out of order because the server is returning them out of order

@nrshapiro I remember you saying you added Roman numeral sorting to the server response as you have access to the decimal numbers there

If not, I can sort it on the client side. Let me know

[nrshapiro]

@jordanallen-dev

That table of contents is "corrupted" by ghost instances which are removed from the file system but remain in Solr during updates (as opposed to full rebuilds). Psychoanalytic-Electronic-Publishing/OpenPubArchive-Content-Server#181

I keep deleting them but they keep coming back to haunt us since we have not done a full rebuild.

Making opasDataCleaner part of the run as described there is an important fix because otherwise, we keep getting those ghost instances reintroduced to stage.

The server is sorting as you recalled, the sorting is correct, but looks wrong because of the ghosts and because the P0000 instance reports its first page number as i, even though it's not declared as a PR0001 instance. But that's ok.

[SophieMBennett]

PI - PI_43_3_April_2023
https://pep-web.org/browse/PI/volumes/43

@jordanallen-dev

Adding here as it seemed to relate to the issues in this thread.

In this instance, the use of a roman numeral in a later issue of the same volume (issue 3) has led to that issue being placed at the very top of the TOC and thus ordered incorrectly. I'm guessing that this might be a platform related issue rather than XML/keying?

cc. @ocappello

[SophieMBennett]

@jordanallen-dev We were discussing roman numerals and TOC ordering earlier this week - there is a recent instance I've been meaning to ask you about, see above (Jun 12 comment)...where the use of a roman numeral in a particular journal issue takes the whole Issue to the top of the list of issues in the TOC taking it out of natural sequence. See https://pep-web.org/browse/PI/volumes/43

The coding for the article in question is below:

I'm wondering if this is something Aptara can fix through amending the coding or whether it would be platform related?