ID passed to Index and Get requests needs to be ServerEncoded

[synhershko]

So URLs and other non standard strings can be used as IDs

[Mpdreamz]

Thanks,

Not all the bits in https://github.com/Mpdreamz/NEST/blob/master/src/Nest/Resolvers/PathResolver.cs are properly passed through Uri.EscapeDataString(), will jump on this tomorrow.

Is this something you need for your talk tomorrow?

[synhershko]

Nah, just something I noticed

[Mpdreamz]

.NET has the nasty "security feature" of forcefully unescaping %2F back to / inside webrequests/uri's.

This behavior can only be controlled from the web|app.config so the the default action NEST takes is to double escape these. This is not ideal as the real Elasticsearch ID will contain a %2F but will for the most part work as expected if you do all your indexing and searching using NEST.

If you set this flag in your application config

<uri>
  <schemeSettings>
    <add name="http" genericUriParserOptions="DontUnescapePathDotsAndSlashes" />
  </schemeSettings>
</uri>

You can instruct NEST to no longer double escape %2F using SetDontDoubleEscapePathDotsAndSlashes() and all should be well.

(more info see: http://msdn.microsoft.com/en-us/library/ee656542%28v=vs.100%29.aspx and http://mikehadlow.blogspot.nl/2011/08/how-to-stop-systemuri-un-escaping.html)

[synhershko]

Do you have to use Uris? Haven't checked, but does HttpWebRequest.Create("http://google.com") convert the string URL to a Uri object?

[synhershko]

Best if Uri can be avoided, also see http://ayende.com/blog/4422/is-select-system-uri-broken