Message198227
| Author |
r.david.murray |
| Recipients |
chortos, petri.lehtinen, python-dev, r.david.murray, serhiy.storchaka, terry.reedy, vstinner |
| Date |
2013年09月21日.22:27:17 |
| SpamBayes Score |
-1.0 |
| Marked as misclassified |
Yes |
| Message-id |
<1379802437.7.0.574325556181.issue12085@psf.upfronthosting.co.za> |
| In-reply-to |
| Content |
__del__ methods are in general tricky because they are in the general case run asynchronously. Therefore any proposal to "attach" the message to another message is a non-starter.
If a __del__ method depends on attributes set in the __init__, then the programmer needs to decide if they want to handle the possibility of __init__ failing, and therefore __del__ running without __init__ having completed. For the stdlib, I think I'd lean toward handling such cases, in which case IMO the Pythonic thing to do is indeed to have a class attribute to provide the pre-__init__ default. |
|