DOC: Remove Series.view(dtype=None) always trying float64

[jorisvandenbossche]

xref https://github.com/pandas-dev/pandas/pull/20220/files#r173655704

Currently, Series.view() is actually equivalent to Series.view('float64') (because that is the default dtype in the Series constructor) which often does not make sense. Not too important (as there are not many reasons to use this function in pandas), but I think we should fix to either a) disallow dtype=None or either b) if dtype is None to set dtype = self.dtype (behaviour of numpy if no dtype is specified)

[jreback]

ok, this is fine, though we should simply deprecate this.

[villasv]

I don't think this function is going to be used a lot, but I wouldn't mind leaving it there instead of deprecating it. I'm specially in favor of the second solutution (if dtype is None to set dtype = self.dtype)

[jreback]

@villasv see my comment son the other issue
this is pretty useful in the current memory model, causes bugs, is not tested and you can simply do this with .astype or use a numpy view directly if you want

having extra functions is plain confusing and just adds technical debt

[villasv]

Sound arguments, I'm in favor of keeping it but not specially attached to it. I won't miss it if it's gone. I'll try to keep discussions in the other issue only.

[mroeschke]

Looks like this respects the original dtype now, so maybe just need to update the docs

[jorisvandenbossche]

We can indeed certainly fix the docs, but IMO we should just deprecate it as well -> #20251

[jorisvandenbossche]

The view() method has been deprecated now, so going to close this issue.