Deleting `Function.Bijection` and friends?

[JacquesCarette]

This was deprecated in v2.0 from December 2023. This is a non-trivial source of {-# OPTIONS --warning=noUserWarning #-} pragmas in stdlib.

A proper cleanup in v3.0 really needs to get rid of these entirely.

[jamesmckinna]

Before or after (any potential) merge of #2583?

UPDATED I saw this before I saw #3034 so I perhaps misunderstood your point @JacquesCarette but I think I strongly agree. Not sure whether #3034 tackles this?

[MatthewDaggitt]

Agreed! #3034 doesn't tackle it as it only handles 1.X deprecations.

[JacquesCarette]

Completely separate things indeed. #3034 is completely uncontroversial. This is an issue because there is mild controversy wrt how much v3.0 is allowed to break. I didn't want to plunge ahead only to have to revert, even though I think this is a big cleanup that we've been waiting a long time to "finish off".

[jamesmckinna]

I agree that v2.0-era deprecations were already the fruit of much re-thinking of the library design, and with it a big commitment to the deprecation/backwards-compatibility discipline.

So in that sense, we perhaps should have regarded it as 'v1.999...' and started the v2.x cycle with v2.1 (eerily reminiscent of sterile discussions about when the millennium took place etc.), but I'm OK with the cruft for the time being.

I think the issue with warning suppression is actually orthogonal. But I need to do more research to confirm that first.

[JacquesCarette]

Happy to treat deprecated Function.* as v1.99 era.

I agree that removing now-useless warning suppression pragmas should be done separately.

[jamesmckinna]

Happy to treat deprecated Function.* as v1.99 era.

See my most recent comments on #3034 ...