On Thu, Dec 14, 2000 at 11:18:31AM +0200, Sheldon Hearn wrote:
> On Wed, 13 Dec 2000 23:19:36 PST, "Crist J. Clark" wrote:
> > I had this problem this weekend. There was a rogue '.depend' file in
> > my _source_ tree. I thought that I had messed something up. Maybe it
> > got there some other way. I think I fixed it with a 'make clean' in
> > the source tree.
> Ha!  That's it.  I wonder how those got there?
> No matter.  We can probably assume that it was due to finger trouble.

OK, I was thinking about this and there is a problem I have with
the fat-finger, pilot-error theory. The only machine that I build
current on mounts the source tree read-only as an NFS. I could not
have accidently left stuff in the source tree. The other issue
is that the problem occured in src/sys/modules tree. I cannot recall
_ever_ building up there. I would only have ever used the buildkernel
target from /usr/src or possibly a 'make' in src/sys/compile/KERNEL.
All of those should be building in different object directories. Could
we have Makefiles breaking occasionally and leaving droppings in the
source tree and we only notice when we get breakage like this pcivar.h

When I presented my earlier solution, I should have mentioned that I
had only cleaned the directory where I had found the '.depend' and '@'
leftovers. I just did a search up my source tree and found,

  [104:/export/current] ls -la src/sys/modules/accf_data/
  total 5
  drwxr-xr-x    2 cvs  cvs   512 Dec 10 00:04 .
  drwxr-xr-x  113 cvs  cvs  2048 Dec  9 13:50 ..
  -rw-r--r--    1 cvs  cvs   548 Nov 17 16:46 .depend
  -rw-r--r--    1 cvs  cvs   179 Jun 19 18:09 Makefile

Anyone else have one of those? And what makes me even more suspicious
is I have that exact same .depend file (same name, not same contents)
in my STABLE tree. It's the only one there too with roughly the same
date (two days later).
Crist J. Clark                           [EMAIL PROTECTED]

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to