Bug #1693
closed
Metacat client library: Static Cookies
Added by Saurabh Garg over 19 years ago.
Updated about 11 years ago.
Description
The HttpMessage code uses static cookies. This is bad and should be changed.
(From bug# 188)
Below given is the initial email from Bill Baker regarding static cookies and
replies from others.
I agree. We noticed this problem a long time ago, but never fixed it.
Sid, could you enter this as a bug and target it at the next major
release? Thanks.
Matt
Jing Tao wrote:
Yep, I agree with you. It is better that cookie is a non-static member.
But I have no idea why it is chosen to be static in HttpMessage class
(utilities module). Do you guys know that?
Jing
Bill Baker wrote:
I guess what I'm really asking is whether cookie should be a
non-static member, since it is treated that way, for example, in
MetacatClient -- the cookie is set, and then a message is sent. Since
there can be multiple client sessions in a JVM simultaneously (for
example, several people logged into the same Metacat HTML UI), that
seems prone to race conditions:
(1) Client A sets session cookie
(2) Client B sets session cookie
(3) Client A sends message, now with B's session cookie
(4) Client B sends message ...
At 16:55 2005-07-28, Jing Tao wrote:
Hi, Bill:
I guess the reason they are static method because cookie is a static
member variable. But it seems better to be non-static methods.
Jing
Bill Baker wrote:
Sid & Jing --
I noticed that HttpMessage.setCookie and .getCookie are static (as
is the cookie member variable) -- should those be non-static instead?
-- Bill
Original Bugzilla ID was 1693
- Status changed from New to Closed
- Assignee changed from Michael Daigle to ben leinfelder
We changed to use standard apache HttpClient and made the cookie sessionId value a per-MetacatClient instance variable.
Also available in: Atom
PDF