More fP sockets pricing insanity

[Fairlight]

On Wed, Nov 29, 2006 at 01:20:41PM -0500, Kenneth Brody, the prominent pundit,
witicized:
> But is has never been the case that a single dclerk would count as
> more than one for anything.

Not -technically-, no.  But as I illustrated, people often think of a
"single" dclerk as their main parent process when in fact they may have
chain-run several children, etc.  The "technical" single dclerk and the
single dclerk that people think about until they run up against it aren't
always the same thing.

> As I recall, under screen(1) it is not possible to determine that
> two processes are actually connected to the same physical terminal.

If 'ps auxwwf' can sort the processes into trees, I see no hinderance to
finding this out.  It can even sort the trees separately for the multiuser
aspect of screen, so you can tell if mantia is using fairlite's session:

mantia    6948  0.2  1.3  5260 1688 ?        S    17:02   0:00      \_ sshd: mantia at pts/13
mantia    6949  0.3  0.8  1952 1096 pts/13   S    17:02   0:00          \_ -tcsh
fairlite  6960  0.1  0.6  1908  772 pts/13   S    17:02   0:00              \_ screen -r fairlite

Sure, the last one is under fairlite's UID, but it's traced under mantia's
process tree.  If 'ps' can find this out, fP should be able to.  It's
almost certainly a simple matter of tracing the PPID's up the chain.

> But it has never been the case that "any time filePro accesses
> 'something', it's considered a 'user'".  You're not talking about
> some daemon starting up lots of dclerks -- you're talking about a
> single dclerk connecting to multiple sockets.

And I agree.  But I was looking at it from the context of, "Yeah, but
system() never should have taken seats and did.  A "single" clerk isn't
always a single seat."  For a change, I looked at it as a layman might.

> I haven't heard anyone else come to your original assumption that
> every time you call SOCKET(), you eat another license.

It's not like it's -documented- up front.  I would know that it doesn't
how, exactly?  It doesn't say it does, it doesn't say it doesn't.  It's
inconclusive by lack of presentation.  That's all I'm trying to say at this
point, the technical issue having been cleared up by you.  I'm saying that
technically you shouldn't have had to take the time to clear it up.

mark->
-- 
Try our new SPF-0 lotion, SunScream[tm].  Get it while it's hot!