re: PF_KEY socket buffer size issue

[matthew green <mrg%eterna.com.au@localhost>]

   On Thu, 11 Sep 2003, matthew green wrote:
   
   >    But sysctl is grosser than /kern. They both are hierarchical ways of
   >    getting info out of the kernel. Everywhere else in the kernel, we
   >    call that a file system. So why not just expose it as such?
   >
   >    sysctl also has two naming schemes (the text one users are used to and 
the
   >    numeric one used internally). Way gross. :-)
   >
   >
   > perhaps.  the point is that sysctl *is* a part of netbsd that isn't 
optional.
   
   So? How will this ever change? If sysctl is used as a reason to not
   improve kernfs, then kernfs will never grow, so sysctl will always be
   needed, so kernfs won't be improved, and so on.


are you suggesting that we make kernfs a *required* part of a running
netbsd system?  if so, then this isn't such a bad idea[*] but i've not 
heard anyone suggest it yet.  only that it's "more clean" than sysctl.

(BTW: a new sysctl implementation that maps names->numbers for you, which
allows for dynamic entries to be added, is on the way i'm told...)



.mrg.


[*] i don't mean "making /kern required part of netbsd" but "if /kern
IS a required part of netbsd".  i'm going to to argue the former, but i'm
not so sure i like it anyway :-)