[API] Some more cleanup ideas

Stephan Bergmann sbergman at redhat.com
Thu Nov 29 04:20:53 PST 2012

On 11/29/2012 01:54 AM, Thorsten Behrens wrote:
> ploughing through offapi & cppumaker (the tool that generates the
> c++ headers), and so far did this:
>   * unpublished accessibility API
>   * renamed XAccessibleEventListener methods
>   * cleansed cppumaker of dead code, RTL_CONSTASCII verbosity, and
>     writing out exception specs

as discussed offline, did you keep the exception specifications as 
SAL_THROW comments?

> There remain the following open questions:
>   * should we keep ~MyClass() {} throw() - or rather have just one
>     single proper virtual ~XInterface() {} throw in the base class
>     (note the missing virtual all over the place) - or bin all
>     exception specs unconditionally?

I would refrain from such a massive change as making ~XInterface 
virtual.  And having nothrow specifications on destructors is fine IMO.

>   * should we kill [oneway] in IDL while at it? IIRC it went away in
>     the bridges anyway, via i#116038 or didn't it?

Yes, [oneway] is effectively completely dead by now (compared to almost 
dead before the binary URP rewrite).

>   * a bit more extreme - kill ::com::sun::star inside office code, by
>     having a ~global "using namespace ::com::sun::star;" or
>     "namespace api = ::com::sun::star;" in e.g. sal/config.h, and
>     writing only _that_ out in the generated c++ headers?
>     (of course keep the old namespace for SDK. And yes, it's no new
>     idea, just a bit less extreme as what's proposed in the wiki)

namespace css = com::sun:.star is the de-facto standard there, so yes, I 
wouldn't mind using that in the generated headers.


More information about the LibreOffice mailing list