Graham,

so if we move these methods to DataObjectImpl you should still be able
to use them by casting your DataObjectPtr to the impl? I think we
should do this in SDO HEAD along with the other 2.1 spec changes.
There should be only a small amount of rework required when you move
the PHP code up to use a 2.1 spec SDO.

Cheers,

On 17/07/07, Graham Charters <[EMAIL PROTECTED]> wrote:
Hi Andy/Pete,

Yes, we do use this method in the PHP SDO code - thanks for remembering us :-)

I think we need to draw a distinction between SDO C++ for applications
and SDO C++ as an embeddable library.  The SDO C++ spec covers the
former and therefore does not talk about get/setUserData.  The library
role of SDO C++ enables us to more easily write native SDO
implementations for other languages (PHP, Ruby, etc...) and is IMO
very important (I guess I would say that :-) ).

Get/setUserData is used by SDO PHP to manage the relationship between
the PHP SDO Objects and C++ SDO Objects.  Earlier versions of the PHP
Extension tried to manage this separately, but this solution was
complex and prone to problems.

I hope this helps.

Regards,

Graham.


On 17/07/07, Pete Robbins <[EMAIL PROTECTED]> wrote:
> Andy, the static code generation was an old experiment and is not
> used.I have been meaning to remove it for some time as it is confusing
> being there.
>
> The get/setUserData was actually put in there at the request of the
> PHP-SDO team. I'm not sure of the details but I think they use this to
> maintain a correlation between the C++ SDO objects and PHP objects???
> This code is not used anywhere within Tuscany SDO (or SCA) code.
>
> This may be a case where a real life application has shown up a
> limitation in the spec and that we should take a proposal to the spec
> group. I'll try and find out how essential this function is and if
> there is another way to work around this to enable a spec compliant
> api in Tuscany
>
> Cheers,
>
> On 17/07/07, Andy Grove <[EMAIL PROTECTED]> wrote:
> > I'm currently looking at some of the issues that my collegaue, Michael
> > Yoder, raised regarding the use of propietary methods in the SDO header
> > files. In particular, I'm looking at the setUserData / getUserData
> > methods in DataObject.h [TUSCANY-1370]. These methods could easily be
> > moved to the DataObjectImpl.h header instead. The methods are only
> > referenced in code generated static client code (generated by
> > DataFactoryImpl::generateInterface).
> >
> > However, I'm nervous about making the change because the current sdotest
> > suite does not excercise the static code generated classes enough to
> > call these methods. For instance, if I change the code generator to call
> > a non-existant method "foo" instead of getUserData or setUserData then
> > the current tests still pass.
> >
> > What is the status of the code generator? The Tuscany web site
> > (http://incubator.apache.org/tuscany/sdo-cpp-faq.html) states that there
> > are no plans to support this feature. Is this correct?
> >
> > Thanks,
> >
> > Andy.
> >
>
>
> --
> Pete
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]




--
Pete

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to