REGR: Series.str.encode("base64")

[simonjayhawkins]

• closes Pandas 1.0.1 gives 1953 TypeError: Cannot use .str.encode with values of inferred dtype 'bytes'. #32048
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[jreback]

not sure if u saw @TomAugspurger comments

you can’t encode bytes only decode them

[simonjayhawkins]

not sure if u saw @TomAugspurger comments

you can’t encode bytes only decode them

see #32048 (comment)

[jreback]

not sure if u saw @TomAugspurger comments
you can’t encode bytes only decode them

see #32048 (comment)

maybe i don’t understand you comment
you can’t encode this in the standard library
why should we allow it

[simonjayhawkins]

we don't use the standard library. we use codecs so maybe we need a deprecation warning.

[jreback]

we don't use the standard library. we use codecs so maybe we need a deprecation warning.

no the point is we shouldn’t do this at all

if you have bytes you can’t encode
i am not sure why that would be an option that is exposed

[simonjayhawkins]

IIUC it's a regression. and if we don't want to allow it we should have a deprecation cycle.

[simonjayhawkins]

@jreback is it worth fixing up the tests and adding a special case for base64 encoding or is this a non-starter.

[jreback]

this was done on purpose. i see no reason to change anything