> > Any objection to doing
> > 
> > .if ${MACHINE_ARCH} != "powerpc"
> > SUBDIR+= except round
> > .endif
> > 
> > in regress/lib/libc/ieeefp/Makefile
> > 
> > for the 1.4.2 release?

That was the first fix I tried, but it just got me to the next bug.
The directory just after libc/ieeefp dies too, and in the same way.

All my 1.4.2 builds on macppc have required changing regress/lib/libc/Makefile
to suppress those two directories; however since two other arch's also turn
these directories off, the actual edit is smaller than you might expect.

> > I mean, "Hello!"...  "make build" really ought to complete
> > without errors.
> 
> Nope. In fact, you'll find comments to that end from me in the releng log.
> :-)
> Bill

Huh? I hope I'm simply failing to comprehend you here...

In my book there are three options for this sort of failure:

  1. fix the problem promptly.
  2. disable it, if it's determined that the test isn't meaningful.
  3. disable it but leave an XXX or similar comment so later grepping sees it.

Leaving something broken such that everyone has to keep editing around it
serves only to waste people's time and makes absolutely no sense to me.
When things are going to be unbuildable for long periods of time they
should happen on side branches, not in the main branch, and DEFINITELY
not on a release branch.

Todd Whitesel
toddpw @ best.com