cursor prefetch with bounded lag

EDIT: nvm, looks like there's really no mechanism in the pgwire protocol that would support this behavior.

Would it at all be possible to support prefetching a batch of up to N rows via a cursor, as compared to waiting for exactly N rows to be available?

In other words, I would like to have a max_lag parameter similar to timeout, but with different semantics. If after max_lag at least 1 row, yet less than prefetch rows are available, do return the available rows.

Current api is:

cursor(query, *args, prefetch=None, timeout=None, record_class=None)

I would like it to be augmented as:

cursor(query, *args, prefetch=None, max_lag=None, timeout=None, record_class=None)

Is there any intrinsic limitation that would make this not worth the effort or simply there has been no interest for anything like that?

Thank you!

Commencez par vérifier la limitation de pgwire indiquée dans l’issue et par examiner l’API de curseur existante décrite dans le rapport. Aucun fichier ni test n’est nommé, et le comportement demandé pourrait ne pas être implémentable ; sa réalisation nécessiterait une conception confirmée et prise en charge par le protocole, ainsi qu’un périmètre de projet correspondant.