Hi Jonathan,

> Attached is the current version of RadioButton Group Name support.

thanks, I'll look into it in more detail later, for now, lemme just
reply to some items in your mail.

> As it happens, your hint about ORadioButtonModel::SetSiblingPropsTo()
> solves the problems I had last time.

That's great!

> 1. The GroupName property has not yet been brought to the ODF file
> format, and thus it currently relies on the default property support.

Usually, I introduce a new attribute in such cases, even if this means
that the documents will not conform to the ODF standard, yet. I think it
is better having a not-yet-compliant-but-readable file format than a
compliant-but-awkward file format.

So, I suggest you add this new attribute (not sure if you need guidance
through xmloff/source/forms ... it should be halfway understandable,
though it's not the most modern code. But this time, it's indeed me and
only me to blame :). Writing the small document which will propose the
addition of this new attribute to an upcoming ODF version can be done
later, the non-existence of this proposal must not hinder us from
integrating the feature.

> 2. GroupName hasn't yet been hooked into any import filters, so Excel
> import won't start using GroupName yet.

That's nothing I can talk about, you would need to direct this to our
Excel filter developer - Daniel Rentz, IIRC. However, I am sure he would
be most grateful for getting this feature, and will surely lend you a
hand in implementing the import/export.
If you happen to have the code in a CWS, then this would ease it - I am
sure we can convince Daniel to write the import/export then.

> 4. About those unit tests... :-)
> 
> First of all, it helps to know that I need to run `dmake` within
> qadevOOo before `dmake` in forms/qa/integration/forms will work.

Uhm. I would have assumed this is built with a regular build ... but
thinking about it, this might in fact not be the case with a vanilla
build environment (remember we still have some slightly different
environment in Hamburg, sadly). Sorry for not telling you, I simply
didn't know.

> Next, the tests work...inconsistently.  Even against an *unmodified*
> OpenOffice.org 2.4.0 installation (as provided by the .tar.gz on the
> openoffice.org site), `dmake run_RadioButtons` fails the *first* time
> it's run within checkThreeGroups -- somehow changing the selected
> RadioButtons results in NO radio buttons being selected (bizarre!).
> This results in one window remaining open after the test.
> 
> Run `dmake run_RadioButtons` a second time (while the remaining window
> from the first run is still open), and all tests work.  Close the window
> "left behind" by the failing run and it will fail in the same spot
> again.

This is a deeper analysis of the problem I did so far :) In fact I
recently noticed this test fails, but didn't know its success depends on
the previous doc window staying open. Indeed, very odd.
I did not yet find the time to investigate this, but for the moment was
happy the second run succeeds ...


> Doubly odd is that occasionally I get a NullPointerException:
> 
>         java.lang.NullPointerException
>         LOG>    at 
> integration.forms.RadioButtons.checkRadio(RadioButtons.java:329)
>         LOG>    at 
> integration.forms.RadioButtons.checkCalcPageSwitch(RadioButtons.java:251)
>         LOG>    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>         LOG>    at 
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
>         LOG>    at 
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
>         LOG>    at java.lang.reflect.Method.invoke(Method.java:585)
>         LOG>    at complexlib.MethodThread.run(Unknown Source)
>         LOG> Finished null
>         ***** State for integration.forms.RadioButtons ******
>         checkCalcPageSwitch - java.lang.NullPointerException
> 
> Not sure what's going on there.

Uh - never encountered this. Looking at the code, this means that the
XAccessible(Value) for the of the radio button control could not be
retrieved. Which is, ehm, quite impossible :-\
It would be interesting to know whether the return value of
accessible.getAccessibleContext() is already null, or only the xValue.


> Again, this happens against both an OpenOffice.org 2.4.0 install and
> against a build with my changes, so I don't think this is due to me.

Agreed - as said, I know the first problem, and the second doesn't
sounds like your changes have any chance of causing it.

>> - m_sGroupName must also be initialized in the "cloning" ctor of
>>   ORadioButtonModel
> 
> Is this "cloning" ctor the ORadioButtonModel(const ORadioButtonModel*
> _pOriginal, const Reference<XMultiServiceFactory>& _rxFactory )

Yes.



Okay, more (perhaps) when I looked at your patch ...

Ciao
Frank

-- 
- Frank Schönheit, Software Engineer         [EMAIL PROTECTED] -
- Sun Microsystems                      http://www.sun.com/staroffice -
- OpenOffice.org Base                       http://dba.openoffice.org -
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

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

Reply via email to