Improve punycode handling with libidn #63
Conversation
Should be somewhat more performant as a regex check is cheaper than PublicSuffix´s parse. It should also help with avoiding running into exceptions in PublicSuffix´s parse (I am guessing a URL with with a bad scheme is more likely to have a “bad” domain).
Fix #48
|
@dentarg think this a viable solution? |
|
Yes, sure |
|
I wonder if http://packages.ubuntu.com/precise/idn vs http://packages.ubuntu.com/precise/libidn11 |
|
Yeah, I wrote about it in the issue #48 (comment) |
|
My guess is that "idn" is just a "wrapper package", i.e. it does nothing except requiring the other package (libidn) |
|
I also wonder what idn2 and IDNA2008 is, seen on http://packages.ubuntu.com/search?keywords=idn |
Think so too. They both require libc6, so I think just libidn11 is fine. I checked one of our app-servers and it (libidn11) was already installed, so I think this will work just fine. |
|
Yeah, but I think we should add it to ansible playbooks anyways |
👍 |
Fixed in https://github.com/twingly/ansible/commit/c8d999f8464fc7a81a51d18d54c690069a7719fc |
|
I think easiest path is to merge this, release bugfix release and try it out on a Heroku app that sees a lot of urls (Twitter?) and see how it goes. |
Yeah, but I think that's "nice to have knowledge", I think we should proceed with what we have going now. |
I think it will work fine.
|
Yes, I agree with all comments above |
|
Great! |
Improve punycode handling with libidn
Requires libidn.
We need to verify that it works in the current locations:
Heroku (will try in production)Datacenter (will try in production)Fixes #48.