[web-devel] unicode domain and path

Sebastiaan Visser haskell at fvisser.nl
Tue Jun 14 18:13:43 CEST 2011


Oh, forgot to mention, you're right.

These functions should only be used on individual path segments and the individual query string, not on the URI in its entire!

Note that there is no real specification describing how to use unicode paths and query strings. Luckily most browser act the same, but, as far as I can remember, clients are free to come up with different encodings.

Thanks,
Sebastiaan

On Jun 14, 2011, at 6:04 PM, Michael Snoyman wrote:
> Hi Sebastiaan,
> 
> That code doesn't seem to take into account the differences between
> query string and path info for which characters are escaped.
> 
> Michael
> 
> On Tue, Jun 14, 2011 at 7:01 PM, Sebastiaan Visser <haskell at fvisser.nl> wrote:
>> Jeremy,
>> 
>> For proper URI encoding and decoding you can use our uri-encode[1] package. It builds on top of Network.URI.
>> 
>> Unfortunately only available via github now, soon also available on hackage.
>> 
>> Cheers,
>> Sebastiaan
>> 
>> [1] https://github.com/silkapp/uri-encode/blob/master/src/Network/URI/Encode.hs
>> 
>> On Jun 14, 2011, at 5:23 AM, Jeremy Hughes wrote:
>>> Hi,
>>> 
>>> The type of both host and path fields of http-enumerator's Request
>>> type is ByteString/Ascii. Does this mean using unicode URLs vai
>>> encodeUtf8 is a no go?
>>> 
>>> J




More information about the web-devel mailing list