Blue Swirl <blauwir...@gmail.com> writes: > On Tue, Jun 15, 2010 at 8:17 AM, Markus Armbruster <arm...@redhat.com> wrote: >> Blue Swirl <blauwir...@gmail.com> writes: >> >>> On Mon, Jun 14, 2010 at 9:36 AM, Markus Armbruster <arm...@redhat.com> >>> wrote: >>>> Blue Swirl <blauwir...@gmail.com> writes: >>>> >>>>> Make APICState completely private to apic.c by using DeviceState >>>>> in external APIs. >>>> >>>> Could you explain why this is an improvement? >>> >>> Outside of apic.c, there is no need to access APICState fields so we >>> can remove that privilege. We can move the device instantiation to the >>> board level where it belongs. >> >> Moving the definition of struct APICState into apic.c is a clear win. >> But what does widening argument types from APICState to DeviceState >> accomplish? The compiler won't be able to catch certain stupid type >> errors anymore; what do we gain for that loss? > > More beautiful architecture. This is how for example Sparc devices > work: there are almost no global functions, all but a few are static.
I'd take the static type checking any day. But it's your funeral, I guess :)
