On Monday 14 January 2002 13:56, you wrote:
> On Mon, Jan 14, 2002 at 01:24:28PM -0500, Gianni Johansson wrote:
> > On Monday 14 January 2002 10:42,  Oskar wrote:
>
> < >
>
> > Oskar, what about fixing this:
> > http://sourceforge.net/tracker/index.php?func=detail&aid=459779&group_id=
> >978&atid=100978
>
> What is so buggy about that? It seems the client worked correctly to me.

Scenario:
User one inserts a jpeg of hops (>32k) under a CHK.
User two  who has never met user one and doesn't even
know that user one exists naively tries to insert the same jpeg of
hops under the key KSK at cool_image_of_freenet_bunny.jpg.  Their
insert attempt fails with the message:

freenet:CHK at KTqMwNk4jL7mqSOqrxq~FtOJ84kKAwI,Y87NDHa3OfrWyk2y33h-Aw' 
already exists in Freenet. (Key Collision)
Request failed

Do you expect user two to be anything other than disgusted and dismayed?  Do 
you expect AutoClient users to be impressed with the fact that they have to 
complicate their code to handle an extra case which serves no purpose?

There is no reason to spuriously fail, when you could easily create the KSK 
redirect to the existing CHK.

>
> > Here are some other ones that keep Java programmers from taking our
> > client interfaces seriously:
> >
> > http://sourceforge.net/tracker/index.php?func=detail&aid=493168&group_id=
> >978&atid=100978
> > http://sourceforge.net/tracker/index.php?func=detail&aid=493174&group_id=
> >978&atid=100978
>
> You submitted those, you could just as well have taken it up here...

I have.  My original implementation of FCPClient had reasonable cancel() 
semantics and didn't write to buckets after cancel, but Tavin backed it out 
and changed the Client interface so that it is impossible to know when 
resources used by a request are released after cancel().  I got shouted down 
in the resulting debate.  Though the issues were never resolved.


--gj

-- 
Freesites
(0.3) freenet:MSK at SSK@enI8YFo3gj8UVh-Au0HpKMftf6QQAgE/homepage//
(0.4) freenet:SSK at npfV5XQijFkF6sXZvuO0o~kG4wEPAgM/homepage//

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to