Hi!


> > Hm.. the SessionInputStream should take care of that. SSCI should only
have
> > transient state so it is ok to store it. When we activate a session with
SIS
> > there is code in there to detect a SSCI and replace it with the real
one.
> > Doesn't that work you say? It should.
>
> Excellent!  The only thing we have to do then is skip the sessionContext
> in the passivation. (right now the passivation tries to read the field -
> and fails.)
>
> Done in CVS.

I'm afraid you misunderstood me. The deserialization process *replaces* old
ctx's but they need to be in there in the first place so that there is
something to replace. The above won't work.

Exactly why can't we serialize the ctx? As far as I can recall there is no
state in it.

(sorry if you've discussed this already, I have missed that in that case)

/Rickard




Reply via email to