RFC: remove the export for Base.require

[vtjnash]

this is in preparation for also removing require(::String) during the incremental compilation work (#8745). otherwise it is confusing to the system whether or not it should be using precompiled versions to satisfy this function.

based on a quick grep of my local packages, they should generally have been doing one of (a) not calling require at all (b) using import X instead or (c) using conditional modules or (d) using import followed by a relative using statement.

also, it's nice to actually be eliminating one of the include/require/using/import set.

[ssfrr]

also, it's nice to actually be eliminating one of the include/require/using/import set.

👍 👍

[timholy]

👍, but can we please do this via a deprecation warning?

[mlubin]

Yes, please add a deprecation

[IainNZ]

OTOH, I'm not sure there is a "legitimate" usage of require on 0.3 code, and it should (I think) be easy to fix (that is, no harder than other things we've muddled with). So if having require still around but deprecated is a blocker for something for 0.4 (like package precomp), I don't think we should feel bad for making it a breaking change.

[JeffBezanson]

Yes, let's just deprecate the function.

[timholy]

One thing to check: programmatic usage of require in JLD, where the module that needs to be included is encoded as a string in an HDF5 file. What is the recommended way forward here?

[vtjnash]

merged in #8745