Hi, sorry for the delayed/missing response, the last few days have been kind of hectic 
(lots of mail).
We are working on improving the situation.

(response follows below)

----- Original Message ----- 
From: Jonas Larsson <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, October 20, 1999 12:15 PM
Subject: Bug in EJB and JSP implementation!


> Hi,
> I've downloaded the core of orionserver version 0.7.7 to solve some of my
> prevoius EJB errors. It solved my previous errors but i've got some new bean
> managed persistence EJB errors:
> 
> 1. With the orion implementation today a single argument function of
> ejbCreate MUST exist. If you for example only implement ejbCreate( long lID,
> double balance ) (and the required interface funktions in the home
> interface) the server throws an error that ejbCreate isn't implemented.
This is a possible bug in the 0.7.7 implementation, we where unable to reproduce it in 
0.7.8
so I suggest you give that version a shot, if the problem persists then let us know.

> 2. OK.. If you then implement a single argument function of ejbCreate orion
> throws an error when it tries to compile it's selfcreated wrapperclasses for
> my EJB:s. It complains that a variable of the name __id dosn't exist. I
> think that some garbage code is present here. I've been investigating the
> generated .java wrapper classes and discoverd that a variable named  id is
> also referenced to.
Yes, another 0.7.7 bug (fixed).
 
> Here I've got some container managed persistence EJB errors using Db2 as the
> persistance store:
> 1. The automatic creation of tables in the persistance storage throws an
> exception about creating the primary key column unique primary key but
> without NOT NULL wich is required for unique indexed fields in DB2.
This doesnt sit right with some other databases - SQL has a fragile syntax  and
lots of inconsistencies between db's, we aim to fix this by providing pluggable type
mappings etc. For now you'll just have to run the query (create table) manually
(by cutting and pasting and changing the needed parts).

> Finally a little jsp error:
> 1. I've been using the <jsp:forward> tag and i've encountered the following
> problem with the jsp:useBean tag:
> err.jsp:
> <jsp:useBean scope="page" id="bean" class="octaglen.Bean1" />
> <html>
> <body>
> <jsp:forward page="err2.jsp" />
> </body>
> </html>
> 
> err2.jsp:
> <jsp:useBean scope="page" id="bean" class="octaglen.Bean2" />
> <html>
> <body>
> </body>
> </html>
> 
> When i try to access err.jsp i get a ClassCastException because the usebean
> (with scope page or request ) in page err.jsp have the same id as err2.jsp.
> Is it really supposed to be like that (I think not!). The exact exception
> comes here:
> 
> 500 Internal Server Error
> java.lang.ClassCastException: octaglen.Bean1
> at __jspPage29._jspService(__jspPage29.java:22)
> at com.evermind.server.http.EvermindHttpJspPage.service(JAX, Compiled Code)
> at com.evermind.server.http.HttpApplication.ln(JAX, Compiled Code)
> at com.evermind.server.http.JSPServlet.service(JAX, Compiled Code)
> at cs.iq(JAX, Compiled Code)
> at cs.forward(JAX, Compiled Code)
> at __jspPage28._jspService(__jspPage28.java:32)
> at com.evermind.server.http.EvermindHttpJspPage.service(JAX, Compiled Code)
> at com.evermind.server.http.HttpApplication.ln(JAX, Compiled Code)
> at com.evermind.server.http.JSPServlet.service(JAX, Compiled Code)
> at cs.iq(JAX, Compiled Code)
> at cs.forward(JAX, Compiled Code)
> at c2.kk(JAX, Compiled Code)
> at c.run(JAX, Compiled Code)
> 
If you get this error when using the page scope it is indeed a bug (we where unable to 
reproduce it though), if
you on the other hand use the request scope then this behaviour is fully compliant 
with the spec (expected behaviour),
both the pages are processed in the same request so the var is already defined for the 
scope by the first page by the
time the second page is invoked.

 
> Questions:
> 1. Will you support clustering of EJB:s in version 1.0 of the server?
No, probably not, we will try to get 1.0 out there in a reasonable timeframe instead.

> 2. If not will you ever support clustering of EJB:s?
Yes, it is definitely on the schedule, perhaps version 1.1/1.2.

> 3. Is there any way to contact the guys at orion without using this list?
Yes, [EMAIL PROTECTED] is good for general bug-reports.
 
> I like your kickass product alot.
Thank you, we appreciate it, have a nice day! :)

/Magnus Stenman, the Orion team

> /Jonas - Keeps my hope alive :)
> 

Reply via email to