axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steve Loughran <ste...@apache.org>
Subject Re: [jira] Created: (AXIS-1596) enum of QNames generates meaningless names
Date Wed, 20 Oct 2004 15:30:52 GMT

On Tue, 19 Oct 2004 15:44:52 +0530, Venkat Reddy <vreddyp@gmail.com> wrote:
 > Steve,
 >
 > I thought it is you who suggested this fix. I knew that such fix is a
 > deviation from the spec, and thats why i asked for further
 > confirmation.
 >
 > Anyway, can we then close this issue?
 >
 > -- Venkat
 >
 >

Here is the problem as I see it.
(1) xsd:enum  to java static instances is broken, because it is too easy 
to get value1, value2 names
(2) we have to remain compliant with that to pass the tests.
(3) the only way to fix JAXRPC is to change the spec.

I have a cunning plan.

(a) we generate value1, value2 bindings
(b) we also generate readable bindings 
_enum_value_all_funny_things_are_underscores
(c) in a miracle of static initialisation we bind the two together such 
that :

  value1== _enum_value_all_funny_things_are_underscores

JAX-RPC tests will pass, code that uses value1, value2 wont break. but 
people can also use sensible names too.

now, is that both devious and workable?


 >
 >
 > On Tue, 12 Oct 2004 17:54:23 +0100, Steve Loughran 
<stevel@apache.org> wrote:
 > >
 > > On Tue, 12 Oct 2004 13:21:06 +0530, Venkat Reddy 
<vreddyp@gmail.com> wrote:
 > > > Dims,
 > > >
 > > > We are on it. Ashu and I have posted comments. We can implement the
 > > > suggestion from Steve, to replace illegal chars with an underscore,
 > > > though it could mean a minor deviation from JAXRPC spec. We need
 > > > confirmation that its fine to go ahead.
 > > >
 > > >  -- Venkat
 > > >
 > >
 > > I'd say its quite a major deviation, to the extent that our generated
 > > stubs would be different from anybody elses. Does the TCK test this?
 > >
 > > -steve
 > >
 >

Mime
View raw message