Message209954
| Author | scoder |
|---|---|
| Recipients | benjamin.peterson, eric.araujo, larry, ncoghlan, pitrou, python-dev, scoder, terry.reedy, yselivanov |
| Date | 2014-02-02.07:01:36 |
| SpamBayes Score | -1.0 |
| Marked as misclassified | Yes |
| Message-id | <52EDED4F.7010806@behnel.de> |
| In-reply-to | <1391308363.08.0.843068039753.issue17159@psf.upfronthosting.co.za> |
| Content | |
|---|---|
I tried the third patch and it works, but when I write this into a docstring:
def func(x, *, y=None):
"""sig=(a,b)"""
then it fails to extract the signature again and returns (a,b) instead.
I also tried putting in some math term as (non-signature) documentation:
def sig(a, b):
"""sig=(a*b)"""
and it raises a ValueError telling me that the signature is invalid.
So, I still think it should prefer a full function interface over the risky
"__text_signature__" quirk whenever it can.
I also think it shouldn't raise an error if the docstring isn't something
that it can parse, but that's not part of this ticket, I guess. |
|
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2014-02-02 07:01:36 | scoder | set | recipients: + scoder, terry.reedy, ncoghlan, pitrou, larry, benjamin.peterson, eric.araujo, python-dev, yselivanov |
| 2014-02-02 07:01:36 | scoder | link | issue17159 messages |
| 2014-02-02 07:01:36 | scoder | create | |