Tim Marsland writes:
> * James Carlson <james.d.carlson at Sun.COM> [2007-06-01 09:34]:
> > If you later find that you need to raise the commitment level (perhaps
> > because it turns out to be a more common problem than originally
> > anticipated), then you end up having to do extra work to scrub away
> > the commitment level from the variable name, and potentially issuing
> > either patches or confusing documentation for it.
> 
> Aliases just aren't that hard.

They're not.  But patches are.

The broader issue is whether the commitment level ought to leak into
the name and, if so, how much.  We haven't done this in the past, so I
think a change like this would be at best confusing at best.  Does it
mean that all non-"__"-prefixed tunables are now supposed to be
public, but that we merely forgot to write man pages?  Who rototills
the code to "fix' this problem?

I suppose that I just don't see a substantial benefit to carving out a
namespace in driver.conf for private tunables.  I think that at the
point where you're hacking at those files, you're already far enough
off in the weeds that carefully-controlled stability levels are of
little consequence.  In other words, we need better tools and
techniques than just "vi /kernel/drv/foo.conf".

> > Though I can see some value to it (easy to grep for "bad" usages over
> > all drivers), I think I'd rather have that sort of meta information in
> > the man page than in the code.
> 
> But if it's something that doesn't have a manpage entry because its
> committment level and expectation-of-usage is so low, it seems a bit
> odd to create an entry simply so we can say "don't rely on this," on the
> offchance that we might want to raise the committment level later??

I don't follow.

I *think* you're objecting to the draft man page entry for the
driver.conf tunable included with this project's case materials.  If
so, then reducing the commitment level means that we would *NOT* ship
a decoy man page -- that text just remains as reference in the ARC
case.

> We should have a manpage entry when we raise the committment level.

Indeed.  I would object to shipping any man page for something that's
deliberately kept private because we don't think it's necessary.

-- 
James Carlson, Solaris Networking              <james.d.carlson at sun.com>
Sun Microsystems / 1 Network Drive         71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677

Reply via email to