On Mon, 25 Feb 2002, Aaron Bannert wrote:
> On Mon, Feb 25, 2002 at 04:58:50PM -0800, Brian Pane wrote: > > > ...or declaring perchild > > experimental-use-only (and thus not critical path for GA) and then > > doing a rewrite at any time. > > When we voted on whether perchild was a showstopper or not I considered > that to be equivalent to "perchild is experimental, use at your own risk". > > The real question is: How do we make it very obvious that perchild is > experimental? A big red WARNING somewhere? Perhaps some code in > configure.in to detect --with-mpm=perchild? That's what documentation is for. If people are going to switch to non-default mpms without reading the documentation, they get what is coming to them. And by the way, +1 on fixing perchild now. I think Ryan should feel free to do anything he wants to improve the mpm, as long as it doesn't involve major mucking with non-mpm code. The only thing that could break is perchild, and perchild is broken anyway. Joshua.