Well, he didn't sway me as much. :-)

How about the rule: "the service implementation name ends with the string 
'ServiceImpl'".

I also am not a fan of having 2 "Impl" files.  But I don't have any really good 
replacement names off the top of my head.  Can we change one of the other file names 
to be called "Interface" and drop the Impl from its corresponding concrete java class?

--
Tom Jordahl


-----Original Message-----
From: Russell Butek [mailto:[EMAIL PROTECTED]]
Sent: Monday, February 04, 2002 5:08 PM
To: [EMAIL PROTECTED]
Subject: RE: Service interface


Tom, I think Richard's swayed me.  I don't think it's a good idea to make
the rules complex.  It should simply be "the service implementation is
named <serviceName>ServiceImpl."  Period.  If we add "unless <serviceName>
ends in Service, in which case it is <serviceName>Impl" then we lose
clarity.

With that said, I still don't particularly like "ServiceImpl".  We happen
to call both the implementation template generated for a binding an "Impl"
and the concrete service implementation an "Impl".  If we could call one or
the other something else I'd be happy.  Any ideas?

binding impl:  <bindingName>Template
service impl:  <serviceName>Impl

binding impl:  <bindingName>Impl
service impl:  Concrete<serviceName>

binding impl:  <bindingName>Impl
service impl:  <serviceName>Factory

Russell Butek
[EMAIL PROTECTED]


Richard Sitze/Charlotte/IBM@IBMUS on 02/04/2002 09:39:46 AM

Please respond to [EMAIL PROTECTED]

To:   [EMAIL PROTECTED]
cc:
Subject:  RE: Service interface



It's generated code.  We are use to seeing "stupid names" coming from
generated code :-).

Seriously, when trying to figure out what the code generators are doing,
minimizing the variences in generated names seems like the best choice.   A
"stupid" artificial name, yet no question on the intent or lineage of the
name.

<ras>

*******************************************
Richard A. Sitze            [EMAIL PROTECTED]




                      Tom Jordahl
                      <tomj@macromedia         To:
"'[EMAIL PROTECTED]'"
                      .com>                    <[EMAIL PROTECTED]>
                                               cc:
                      02/04/2002 09:20         Subject: RE: Service
interface
                      AM
                      Please respond
                      to axis-dev







Here is a suggestion: Check for "Service" at the end of the name, and don't
append it if it's already there.  This prevents the stupid name. Still
avoids name clashes.

--
Tom Jordahl


-----Original Message-----
From: Russell Butek [mailto:[EMAIL PROTECTED]]
Sent: Monday, February 04, 2002 10:09 AM
To: [EMAIL PROTECTED]
Subject: Service interface


JAX-RPC 0.6 discussed generating a service interface and a service
implementation.  The interface was optional.  In the interests of
minimizing generated files, AXIS opted to NOT generate the interface.

In JAX-RPC 0.7 the service interface is no longer optional.  If the service
is named X, the service interface is named X.  That's the name we'd been
using for the implementation.  I would like to rename the implementation to
XServiceImpl.  I will not simply rename it XImpl because I've occasionally
seen a binding and a service share the same name and we already generate an
implementation template called <bindingName>Impl.  Of course, most often
the service name is something like XService, so we'd be generating the
rather strangely named class XServiceServiceImpl.  I'd be happy to hear
better suggestions if folks have them.

Russell Butek
[EMAIL PROTECTED]




Reply via email to