Author ncoghlan
Recipients Trundle, belopolsky, brett.cannon, brian.curtin, cool-RR, eric.araujo, eric.snow, giampaolo.rodola, gruszczy, loewis, ncoghlan, nedbat, pitrou, r.david.murray, techtonik, vstinner
Date 2011-12-15.12:13:42
SpamBayes Score 2.46442e-11
Marked as misclassified No
Message-id <1323951223.73.0.949486828541.issue1559549@psf.upfronthosting.co.za>
In-reply-to
Content
The keyword-only idea is a backwards compatibility hack we discussed at the PyCon US sprints because ImportError currently accepts an arbitrary number of arguments:

>>> raise ImportError(1, 2, 3)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: (1, 2, 3)

We don't really want to be claiming that the module name is some strange value due to existing code that passes multiple arguments, hence the suggestion to make this a keyword-only argument.

Regarding import.c, I think Brian misinterpreted Brett's last message. import.c absolutely *should* be modified by this patch, since it's still the default import implementation for the moment. Brett's comment was just to say that *he* wasn't going to do that part, since his aim with the importlib bootstrapping exercise is to nuke most of import.c from orbit :)
History
Date User Action Args
2011-12-15 12:13:43ncoghlansetrecipients: + ncoghlan, loewis, brett.cannon, belopolsky, pitrou, vstinner, techtonik, giampaolo.rodola, nedbat, eric.araujo, r.david.murray, Trundle, brian.curtin, gruszczy, cool-RR, eric.snow
2011-12-15 12:13:43ncoghlansetmessageid: <1323951223.73.0.949486828541.issue1559549@psf.upfronthosting.co.za>
2011-12-15 12:13:43ncoghlanlinkissue1559549 messages
2011-12-15 12:13:42ncoghlancreate