Questions for deployments of GNOME

Questions for deployments of GNOME

I consider the Network Object Model to have died with the near-death of Bonobo and its support libraries.

KDE might use a not-very-network-friendly IPC mechanism, but KParts is nonetheless a hell of a lot closer to the way to do componented software than is GNOME. (It's odd, really: the GNOME devs laud MS's software design principles, but it seems to me that they applaud and emulate mostly the parts which are bad: e.g. the interview I read last year which applauded the idea of moving everything into non-command-line-accessible APIs for this and that, with never a consideration that command-line access might be useful (disclaimer: weak memory and hay fever drugs may have totally distorted this; if it is nonsense, my apologies). Yet MS are heavy users of components... and GNOME isn't. Strange. Of course KDE provides command-line access to its IPC mechanism via kdcop; if GNOME provides such access to CORBA, I've never found it.)

(And yes, security holes in heavily-used components are bad things, as the holes in the HTML renderer in IE make clear. But we're not safe from that, as just using libraries is enough to trigger it: a hole in Gecko or glibc or libX11 could be just as dangerous if it were remotely exploitable. The key isn't `don't write components, they're insecure': it's `don't write insecure crap and then make it a component'.)