Message165046
| Author |
orsenthil |
| Recipients |
antlong, christian.heimes, eric.araujo, jerub, orsenthil, pitrou, python-dev, rosslagerwall, wichert |
| Date |
2012年07月08日.23:49:33 |
| SpamBayes Score |
-1.0 |
| Marked as misclassified |
Yes |
| Message-id |
<CAPOVWOTvn-LANn+ScqZnnoRD2fsr_zo1rr-BM4de0+pWnQBopA@mail.gmail.com> |
| In-reply-to |
<1341739832.34.0.366414803172.issue14826@psf.upfronthosting.co.za> |
| Content |
On Sun, Jul 8, 2012 at 2:30 AM, Antoine Pitrou <report@bugs.python.org> wrote:
>
> Senthil, do you read python-dev? I think this change was prematurate from the start (nevermind the fact that you didn't run the test suite before committing).
I thought that the other legacy URLOpen was quoting it correct and
then I wanted to see it can be made consistent.
It did get me thinking that why it was different for so long. I
realize that committing soon was a mistake.
> For example, if you have an URL with a non-ASCII domain name such as "http://وزارة-الأتصالات.مصر/", the domain name should IDNA-encoded, not %-encoded like the rest.
Agreed and understood.
> Furthermore, some people are certainly already quoting their URLs to workaround this issue, so "fixing" it will break their code by double-escaping the URLs. You've got to be more careful.
Oh. yes, the change may break an already quoted URL. I think, I shall
revert this back. |
|