This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author yselivanov
Recipients Yury.Selivanov, larry, skrah, yselivanov
Date 2014-04-30.16:40:58
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1398876058.51.0.45898970343.issue21399@psf.upfronthosting.co.za>
In-reply-to
Content
> In Python2.7, the cls parameter shows up in pydoc:
>
>    frombuf(cls, buf) from __builtin__.type
>        Construct a TarInfo object from a 512 byte string buffer.
>
>
> In 3.5, it doesn't:
>
>    frombuf(buf, encoding, errors) from builtins.type
>        Construct a TarInfo object from a 512 byte bytes object.

Yes, that's a correct behaviour in 3.4 and 3.5. See #20710 for details.

> >>> signature(TarInfo.create_gnu_header)
> <Signature at 0x7f50cf110cf0 "(self, info, encoding, errors)">
> >>> signature(TarInfo.frombuf)
> <Signature at 0x7f50cf11cc88 "(buf, encoding, errors)">

There is no bug here. `TarInfo.create_gnu_header` is an unbound method, that indeed requires first argument 'self'.  `TarInfo.frombuf` is a classmethod, so it doesn't need a 'cls' arg.  Signature, by default and by design, only shows arguments that need to be passed to correctly execute the given callable.

> How about the C docstrings? Can we get "$cls" for classmethods?

Yes, I think it should work.
History
Date User Action Args
2014-04-30 16:40:58yselivanovsetrecipients: + yselivanov, larry, skrah, Yury.Selivanov
2014-04-30 16:40:58yselivanovsetmessageid: <1398876058.51.0.45898970343.issue21399@psf.upfronthosting.co.za>
2014-04-30 16:40:58yselivanovlinkissue21399 messages
2014-04-30 16:40:58yselivanovcreate