Re: [PATCH] rollback of unwarranted change (was: [PATCH] unwarranted

["Roy T. Fielding" <fi...@kiwi.ICS.UCI.EDU>]

I don't mind the veto; the whole point of peer review is to correct
oversights and unwanted changes.

>    You just *don't* go around arbitrarily deleting global cells,
>    particularly ones explicitly *labeled* as being global, and
>    especially not as undocumented riders to other fixes.

I disagree.  It is unsafe for that variable to be changed outside
the configuration, which is why I changed it to an #ifndef.  No
running module has any business mucking with that value, period.
It was my intention to break any such dependency outside the core
distribution before 1.2 was released, because any such dependency
is bad design.  There was nothing arbitrary about it.

If you have a module that needs to selectively turn off IdentityCheck,
then do that in the per-dir config.  If you want to add a configurable
timeout, then code-up a replacement for rfc1413.c.  Do not rely on
broken abstractions for the implementation of features.  The broken
stuff will be removed whenever someone like me needs to squash a bug
in the dark without a flashlight.

....Roy