Re: Bug in setPersistentDomain?

[Andreas Heppel]

On 2002-11-12 13:00:18 +0000 Richard Frith-Macdonald <richard@brainstorm.co.uk> 
wrote:

> On Tuesday, November 12, 2002, at 08:42  am, Andreas Heppel wrote:
>
> > On 2002-11-11 18:46:14 +0000 Richard Frith-Macdonald <richard@brainstorm.co.uk> 
> > wrote:
> >
> > > On Monday, November 11, 2002, at 01:40  pm, Andreas Heppel wrote:
> > > > Hi there,
> > > > I just checked some methods for NSUserDomain and saw that
> > > > - (void) setPersistentDomain:forName: checks the variable _tempDomains whether 
> > > > the requested domainname already exists. Shouldn't this check be made over 
> > > > _persDomains instead? If I get it right _tempDomains holds the volatile ones.
> > > The check shouldn't be there at all (I removed it) and the one in 
> > > setVolatileDomain:forName: was wrong ... I corrected it.
> > OK. But iirc Apple's Foundation manual says that both methods should raise 
> > exceptions if a persistent/volatile domain with the given name already exists, 
> > while the comment in gnustep-base reads 'Replaces the persistent...'. I 
> > actually prefer the latter (and now impelmented :-) version of GNUstep since it 
> > makes more sense to me. But wouldn't that break the *step specs?
>
> Argh ... I checked ... the original implementation was correct according to the 
> OpenStep spec.
> I've reinstated the original tests, so that the code conforms to spec again 
> *and* conforms to
> the MacOS-X documentation.
Is this actually possible? Having checked both specs now (OpenStep _and_ OSX) 
they are completely different in those points. Or am I missing something here? 
I must admit that I missed to read the OpenStep spec earlier and only had the 
OSX one in mind.
Sorry for causing this confusion :-)

> Thinking about it, the OpenStep spec makes more sense than the MacOS-X one ...
> basically it says that you can't have a volatile domain and a persistent domain 
> with the same name,
> wheras MacOS-X says you can't set a volatile domaion unless you remove any old 
> domain of the
> same name first.
I agree. That truely makes sense.
Concerning the OSX specs, is it possible that we do have different issues? The 
document I have here reads that setPersistentDomain:forName: will raise an 
exception if a _persistent_ domain of this name already exists. The same goes 
for setVolatileDomain:forName:. This is complete nonsense to me, because doing 
it like this would mean I can't change a domain without previously removing it. 
Thus my question above how it can be that GNUstep conforms to both.

Andreas

--
Andreas Heppel

Mail: aheppel at web dot de
Home: http://www.andreasheppel.de

Check out GSburn.app - the CD burning frontend for GNUstep
http://gsburn.sourceforge.net