Series.combine and ExtensionArrays

[TomAugspurger]

Working on a PeriodArray. For the first thing I tried, Series.combine doesn't work.

In

pandas/pandas/core/series.py

Lines 2324 to 2330 in 7343fd3

	elif is_extension_array_dtype(self.values):
	# The function can return something of any type, so check
	# if the type is compatible with the calling EA
	try:
	new_values = self._values._from_sequence(new_values)
	except TypeError:
	pass

we catch a TypeError. Why is that just a TypeError, when in principal the EA constructor could throw any kind of exception? In my case a ValueError is ValueError since the freq can't be inferred.

I think that should be except Exception instead. Thoughts @Dr-Irv?

[jorisvandenbossche]

Yes, I think that can be broadened to general exception.

[Dr-Irv]

IMHO, I would just add the errors that you know that could occur. So if ValueError can be raised by ._from_sequence(), then I think you should do except TypeError, ValueError. If some other error is raised, then it might catch another issue in someone's code.

[TomAugspurger]

_from_sequence is completely up to the EA author and could raise anything.

…

On Thu, Sep 27, 2018 at 12:16 PM Dr. Irv ***@***.***> wrote: IMHO, I would just add the errors that you know that could occur. So if ValueError can be raised by ._from_sequence(), then I think you should do except TypeError, ValueError. If some other error is raised, then it might catch another issue in someone's code. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#22850 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIqe1JATKA8XlvqQVREUQ_wuAqkolks5ufQgLgaJpZM4W7zac> .