"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: > It all sounds like a Lisp Machine... And even though I enjoyed your > little story, it has zlich todo with filesystems and user > interactions.
Sure it does. It's about why a filesystem needs to be able to prompt the user, and why RPC should be procedure call. Think of it in those terms. Now that the filesystem is out of the kernel, it's time for it to be allowed to all the things that the kernel can't. I thought I had made sure it could, but I failed to see clearly that it was more than just sheer ability to do things that was needed, but also access to gobs of user context. > I don't know about Unix, but on GNU/Hurd you could suspend the > process, then have some gdb magic so you can restart the process in > question from a previous point (and it would be nice to be able to > patch the process in question with new code, to fix a bug or similar). No, you couldn't, because it's already too late. You can't tell even what process is concerned, and the system is already out of memory and it's too late to go around running other things. >From what I read, Marcus is right that L4 need not have these problems, but Mach does, and the Hurd doesn't yet have the ability to deal even if the kernel were to suddenly develop it. > This is far simpler then having "dialogs poping up", it also begs the > question "where do those dialog windows pop up"; X11? the console? The > Hurd's new-fandangled-windowing-system-that-doesn't-exist-yet? All of > the above? Wherever the current user interaction context says to. _______________________________________________ Bug-hurd mailing list Bug-hurd@gnu.org http://lists.gnu.org/mailman/listinfo/bug-hurd
