>
> On 22 Oct 2013, at 18:27, "[hidden email]" <[hidden email]> wrote:
>
>> Ok, point taken.
>>
>> Still, ZnClient is unusable for the intended purpose as is. No matter what encoding, % or not, Olivier cannot get the URL to work. Couldn't the ZnClient use something like the StampClient does: options? (where prefetch-count can be set for example - I also discovered the little inbox thing after some hair pulling sessions BTW).
>>
>>
>> Phil
>
> But Olivier's problem is that he is talking to a server that does not do proper percent decoding, IMHO. A server parsing a URL should resolve all percent encodings, after parsing using the significant delimiters (/ : ? #).
>
> http://foo.com/abc and http://foo.com/%61%62%63 are the same thing, just like http://foo.com/abc@1 and http://foo.com/abc%401 are - again as far as I understand it.
>
> That being said, I am going to study the specs again and might change what gets encoded or not depending on the URL element, instead of encoding aggressively in the better safe than sorry way.
>
> Olivier's suggestion about safer authorisation parsing does make sense and will get added as well - Thanks Olivier !
>
> Sven
>
> PS: BTW, ZnClient has such an option object already.
>
>