At 03:02 PM 9/14/2002, Ian Holsman wrote:
>On Sat, 14 Sep 2002 05:26:25 -0700, William A. Rowe, Jr. wrote:
>
> > The following patch changes the custom_log init phase to return an error
> > from the config phase, rather than terminating abruptly.
> >
> > It still doesn't seem right. Why are we skipping log intialization on
> > preflight? I understand that the log pipe is expensive, but is there any
> > reason not to run it at preflight and keep that pipe between restarts,
> > just as we do now?
> >
> > Since we don't try opening the output of CustomLog directives on
> > preflight, -t won't warn that things are amis.
>
>+1 for this change to make this release.
>the reason is that it would require a magic # change (and we are already
>doing one for .41, so we might want to avoid doing one for .42)
Thanks for reviewing, but I don't see how an MMN bump is required.
Please explain?
Bill
|