Drop reference to implicit conversion from named to regular tuples

[odersky]

The NamedTuple spec mentioned that .toTuple was inserted implicitly to convert from a named to a regular tuple. But this was not implemented. The sentiment is that this presents an opportunity to get more experience with the feature before deciding whether the conversion is actually beneficial. For now, one can implement the conversion by hand if desired, like this:

given [N <: Tuple, V <: Tuple] => Conversion[NamedTuple.NamedTuple[N, V], V] = _.toTuple

Question: Should we put that conversion somewhere in the stdlib so that one can import it?

[ritschwumm]

since i haven't used named tuples yet, i can only give a gut feeling:

if someone calls toTuple manually it's much easier for me to figure out what's happening when i read her code, so i'd advise against using such a given.

i guess if at some time in the future there is enough evidence the convenience trumps "explicit is better than implicit", it can still be added without breaking existing code.

[EugeneFlesselle]

My understanding was that we do insert toTuple as described in the spec, specifically in adaptToSubType:

scala3/compiler/src/dotty/tools/dotc/typer/Typer.scala

Lines 4644 to 4645 in 0afabd6

	if tree.tpe.widen.isNamedTupleType && pt.derivesFrom(defn.TupleClass) then
	readapt(typed(untpd.Select(untpd.TypedSplice(tree), nme.toTuple)))

Are there cases where we fail to insert toTuple as expected?

[odersky]

Are there cases where we fail to insert toTuple as expected?

There are. See #22028. But since we already have a partial implementation which is postulated by the SIP, it's better to fix the issue as is done in #22028 than backpedalling and removing the functionality. So I am closing this.