The implementation of the RFC2965 spec itself appears to be fine, at
least as a whole.  The only big problem I see is with the integration
into the rest of the code and compatibility with other cookie specs. 
I also agree with Oleg's comments.  getCookieVersion() does not seem
necessary, perhaps a CookieSpec2 interface would be better?  Cookie2
instead of changing Cookie also sounds good to me.

In particular there are some issues with section 9 of the RFC. 
Currently there is no way to support just Set-Cookie2 and only RFC2109
can be used in conjunction with Set-Cookie2.  In addition the use of
the Cookie2 header in the request does not appear to be correct by my
reading.  I also have some questions about what we should do when
sending cookies as a single header and when there are old and new
cookie to be sent.

There are also some smaller things I noticed:
 - HttpMethodBase.buildCookie2RequestHeader() has default access which
does not seem necessary.
 - There are a few System.out.println()s
 - The static vars added to Header and Cookie should be moved to
HttpConstants and RFC2965Spec respectively.

Mike

On 8/26/05, Oleg Kalnichevski <[EMAIL PROTECTED]> wrote:
> Samit,
> 
> These are my first impressions about the overall design. All my
> suggestions are primarily intended to minimize the impact on the
> existing API. They are not necessarily the best ideas from the design
> perspective. However, since we are redesigning 3.0 API anyway, I
> personally do not see a big problem with that.
> 
> (1) We should not change the CookieSpec interface. I am not entirely
> sure CookieSpec#getCookieVersion() is necessary at all. I think the
> CookiePolicy class can be used instead to achieve the same net result. A
> method of the CookiePolicy class that returns a cookie spec given the
> cookie version may well do the job
> 
> (2) Cookie class changed to include RFC2965 specific attributes. I would
> rather extend Cookie class and put all RFC2965 specific attributes into
> a Cookie2 class, if that is feasible.
> 
> (3) We have 39 failing test cases
> 
> Test cases breakage must be fixed before the deadline. Other stuff is
> subjective and thus can be addressed past Sept 1st, provided you agree
> with my points
> 
> Oleg
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to