jed / cookie-node Goto Github PK

For opaque cookies, use symmetric encrpytion instead of base64 of the cookie value/payload.

One doesn't always want to share the contents of the cookie with the user, even if tampering with the value is detectable by an invalid signature.

I suppose one could encrypt beforehand? Hmm, maybe that is a better idea.

Thoughts?

Hi!

Couldn;t you kindly add HttpOnly option?

--Vladimir

Hi Jed

Seems like you wrote the almost exact same method twice :

http.IncomingMessage.prototype._parseCookies

The first one is obsolete and should be deleted.

If you want to set multiple cookies you will see that this is impossible. The problem is described in this post http://caolanmcmahon.com/multiple_set_cookie_headers_in_node_js.html

Its undocumented how to use cookie-node alongside a writeHead call. Only checking the source for cookie-node showed me it was looking for headers as an object -- I was passing headers to writeHead as an array of arrays, which is also supported, but cookie-node doesn't work with that.

Several mandatory fields are currently not included: keywords, maintainers, contributors, bugs, licenses, repositories and dependencies. In addition, some of the optional fields may be relevant.

--source: http://wiki.commonjs.org/wiki/Packages/1.0

I was reading through the code and couldn't understand how processCookie() could work if the cookie.secret isn't set. In processCookie() If the secret isn't set the localSig gets initialised with a new random secret and will fail the test against remoteSig.

..or have I missed something?

heads up, I think for node.js 0.3.3 compatibility line 86 needs to change from

args.push(headers = []);
to
args.push(headers = {});

... like MIT :)