Brian Behlendorf wrote:
>
> On Thu, 17 Jun 1999, Jim Jagielski wrote:
> > Dean Gaudet wrote:
> > >
> > > Why not just remove the stat() altogether? Seems pointless to me... folks
> > > will get various file not found errors in the error_log anyhow if the
> > > docroot doesn't exist; or get ENOTDIR errors if the file isn't a
> > > directory.
> > >
> >
> > I thought about that, but think that keeping it around makes sense
> > for some/most setups because:
> >
> > 1. Right now, if DocumementRoot doesn't exist they get an immediate
> > notice of that when Apache starts. If we change it, then people
> > will assume that all is well and not bother checking error_log
> > until their customers complain. So for people who "depend"
> > on the current behavior, we maintain it.
>
> I think this could be handled by being very clear about the behavior
> change in the CHANGES file. This is pretty minor IMHO, and less
> disruptive than adding another config directive we'll need to support
> indefinitely. =)
>
Whatever.
Why do we bother having stuff in STATUS and "voted" on and then
commited (with a note beforehand "if anyone has problems, let us know")
if a week or so later people will say "you know, we should just do
it differently." Isn't that what STATUS is for?
No matter what, the stat is useful for apache -t. If we want to drop
the directive and have it do the stat for -t then that's fine, but
it sure would be nice if these were brought up beforehand :)
--
===========================================================================
Jim Jagielski ||| jim@jaguNET.com ||| http://www.jaguNET.com/
"That's no ordinary rabbit... that's the most foul,
cruel and bad-tempered rodent you ever laid eyes on"
|