changing the storage model: how hard

[mlinchitstest]

Currently the reference database is pretty useless because (in my experience) it does not work at all once you get to about 500+ bibtex entries. There are two solutions to this: either

• use one bibtex file and cache it in a sqlite database like this old and vulnerable bibtex plugin that no longer works; or
• or have one bibtex entry per page and have the page name be the bibtex key; the plugin currently has to search all the text in the refrence namespace to get the keys. It also has to identify the format: bibtex? note? table? There is no way this can perform well with a large number of references. Furthermore, managing a reference "database" that is just a big glob of text is quite tedious. Obviously, if we're not using an actual database synced to an actual bibtex file (one that can be managed by an external program), it would be more efficient to lookup the citations by the actual file name, since lookup by filename is trivial.

How hard would it be to change the plugin to get this sort of behavior?

[dwp-forge]

This plugin was developed to allow more flexibility in formatting footnotes. Using it with 500+ BibTeX entries is way beyond its scope.

As for how hard that would be, you can try and see for yourself ;)

[mlinchitstest]

i think that just pointing to actual filenames/actual namespaces would solve the problem. how much of the code would need to be changed to that?

[dwp-forge]

It's hard to estimate how big the changes would be. And that's not the measure that has much relevance anyway. It's not about the number of changed lines, it's about knowing which lines to change, and how. The changes should not break existing storage model, and ideally should somehow merge with it. If the current design is done well the changes should be rather local. But I haven't looked in the code for quite some time, so I don't remember anymore. And currently I have no time to do major work on the plugin. But merge requests are welcome.