[WIP] Enhance customization of entry types

[aqurilla]

This fixes #9840 by adding a new data format for custom entry types that supports saving multiple field properties

Mandatory checks

• Change in CHANGELOG.md described in a way that is understandable for the average user (if applicable)
• Tests created for changes (if applicable)
• Manually tested changed features in running JabRef (always required)
• Screenshots added in PR description (for UI changes)
• Checked developer's documentation: Is the information available and up to date? If not, I outlined it in this pull request.
• Checked documentation: Is the information available and up to date? If not, I created an issue at https://github.com/JabRef/user-documentation/issues or, even better, I submitted a pull request to the documentation repository.

[aqurilla]

Hi! Should org.jabref.logic.exporter.MetaDataSerializer#serializeCustomEntryTypes be updated so that custom entry types are always serialized to the v2 format going forward?

[koppor]

Hi! Should org.jabref.logic.exporter.MetaDataSerializer#serializeCustomEntryTypes be updated so that custom entry types are always serialized to the v2 format going forward?

Meta: I miss what you would do :p

You know, I am thinking in MADR, especially using "decision drivers".

Decicion drivers:

• Maintainable solution (code, releases)
• Backward compatibility if possible (some users do not switch JabRef versions - or use two versions in parallel, because newer versions sometimes do not work)

I would vote to always use v2 here - but then include the feature in the next major release (because this breaks compatibility). However, I do not see JabRef 6.0 coming the next weeks.

Thus, use (existing) v1 serialization if single field properties and use v2 if v1 cannot be used. -- For JabRef 7.0 (!) we can remove that logic and always migrate to v2.

[aqurilla]

Same here! 😄

Thankyou for the pointers!
I think we would have to use v2 in all cases where a custom FieldProperty is specified, even if it is just a single field property?

For example, using the person entry type with Name field having a single field property PERSON_NAMES.
Using v1 we would get: jabref-entrytype: person: req[Name] opt[]

Using v2 would retain the field property info: jabref-entrytype-v2: person: req[Name|PERSON_NAMES] opt[]

[koppor]

I think we would have to use v2 in all cases where a custom FieldProperty is specified, even if it is just a single field property?

I fully agree!

What could work is mixing v1 and v2? Only use v2 for those entry types requiring it and write the others as v1. Thus, not all are "lost" when opening with an old version, but only "some".

Thinking aloud: We should not do, because of too many changes in the bib file: write v1 always and v2 *additionally. The reader then needs to prioritize v2 over v1. We could keep that for version 6.0 and remove that for version 6.1.

[aqurilla]

Thinking aloud: We should not do, because of too many changes in the bib file: write v1 always and v2 *additionally. The reader then needs to prioritize v2 over v1. We could keep that for version 6.0 and remove that for version 6.1.

yes that sounds good!
we can write both v1 and v2, and the reader can prioritize v2 if present

[aqurilla]

Hi @koppor, the parser has been updated so that org.jabref.logic.importer.util.MetaDataParser#parseCustomEntryType can parse both v1 and v2 entry types.

Regarding backward compatibility, it looks like even if we write both v1 and v2 it can lead to an incorrect additional entry type when reading using an older version, because the v2 entry type would be parsed incorrectly. Given this, I was wondering if we can write only the v2 version in the Serializer?

[koppor]

Regarding backward compatibility, it looks like even if we write both v1 and v2 it can lead to an incorrect additional entry type when reading using an older version, because the v2 entry type would be parsed incorrectly. Given this, I was wondering if we can write only the v2 version in the Serializer?

Without reading the code, I don't understand. Isn't v2 just ignored in older versions? I thought, unknown meta data keys would be ignored?

Please start to craft an ADR (https://adr.github.io/madr/) to capture pros and cons of each option.