[01:45:22] --- lars.malinowsky has become available
[01:45:29] --- lars.malinowsky has left
[03:52:30] --- Russ has left: Disconnected
[05:04:31] --- meffie has become available
[05:50:15] --- Chris Garrison has become available
[07:24:03] --- steven.jenkins has left
[07:30:37] --- steven.jenkins has become available
[07:44:10] --- deason/gmail has become available
[08:09:50] --- steven.jenkins has left
[08:13:26] --- steven.jenkins has become available
[08:31:25] --- mfelliott has become available
[08:50:49] <shadow@gmail.com/barnowl754E0B64> > It's possible that the scripts get run locally, and

yup
[09:07:47] --- Chris Garrison has left
[09:44:26] --- Russ has become available
[10:31:57] --- Russ has left: Disconnected
[10:43:04] <shadow@gmail.com/barnowl754E0B64> > Next time Derrick pops up

fwiw, i did do this and then forgot to mention it
[12:56:39] --- mfelliott has left
[13:15:48] --- jaltman/FrogsLeap has left: Disconnected
[13:18:06] --- jaltman/FrogsLeap has become available
[13:32:30] <Simon Wilkinson> where this == ?
[13:45:32] --- lars.malinowsky has become available
[13:45:53] --- lars.malinowsky has left
[15:21:33] --- jaltman/FrogsLeap has left: Disconnected
[15:54:35] --- deason/gmail has left
[16:45:55] --- deason/gmail has become available
[17:10:30] <kaduk@mit.edu/barnowl> It looks like Derrick commented on the "FBSD: don't FlushAllVCaches"
change that "the rc file change is missing"?  I think I am confused.
[17:24:30] <shadow@gmail.com/barnowl754E0B64> >  It looks like Derrick commented on the "FBSD: don't
> FlushAllVCaches" change that "the rc file change is missing"?  I think I am
confused.

well, it failed to verify, and that's why...
[17:25:56] <kaduk@mit.edu/barnowl> I'm really confused why it would have, though -- that was a two-line
change to afs_call.c .  How does an rc script come into it?
[17:26:09] <shadow@gmail.com/barnowl754E0B64> make dest failed.
[17:26:15] <shadow@gmail.com/barnowl754E0B64> it builds the whole tree
[17:26:45] <kaduk@mit.edu/barnowl> Ah.
[17:27:24] <kaduk@mit.edu/barnowl> (And we only noticed now because the fbsd buildbot is now building every
change?)
[17:31:42] <shadow@gmail.com/barnowl754E0B64> yup
[17:33:12] <kaduk@mit.edu/barnowl> Nifty.
Do I want to try and get bsd.kmod.mk into 1.6.0?
[17:35:00] <shadow@gmail.com/barnowl754E0B64> well, you might as well start on it
[17:48:48] <kaduk@mit.edu/barnowl> Well, I have a working tree that seems to work.  (But need to roll
back to MakefileProto.in and friends, &c.)
[17:50:18] <kaduk@mit.edu/barnowl> (For values of "seems to work" equal to "produces a libafs.ko")
[17:56:02] <shadow@gmail.com/barnowl754E0B64> push a diff?
[17:57:07] --- rra has left: Disconnected
[18:15:01] --- Russ has become available
[19:26:24] --- jaltman has become available
[19:27:16] --- jaltman has left: Disconnected
[19:27:35] --- jaltman has become available
[19:39:39] --- jaltman has left: Disconnected
[19:45:32] --- jaltman has become available
[19:51:36] --- jaltman has left: Disconnected
[19:56:24] --- jaltman has become available
[19:56:44] <kaduk@mit.edu/barnowl> Hmm, broken fileserver build on master?
[20:03:58] --- jaltman has left: Disconnected
[20:07:12] --- jaltman has become available
[20:11:11] <shadow@gmail.com/barnowl754E0B64> well, i have one from this morning running somewhere, so i'll go with
"no"
[20:12:08] <kaduk@mit.edu/barnowl> I'm tentatively going with stale build products in my tree.
[21:53:44] <kaduk@mit.edu/barnowl> Bah.  (My previous bsd.kmod.mk "success" was spoiled by a spurious
vnode_if.h that had been floating around.)
But, I think that the sourc of my troubles is that our CRULE_NOOPT and
CRULE_OPT are essentially doing 'gcc -c $?', where $? is those
sources which are out-of-date.  The system makefile infrastructure
ends up adding a dependency on ${SRCS} and ${_ILINKS} to (all? many?)
targets.  This results in passing '@ machine x86 vnode_if.h' to the
end of the gcc invocation, and it rightly bails on more than one input
file with -c.
I can work around this by redefining CRULE_{NO,}OPT to use ${?:M*.c} on
my system, that is, those sources which are out-of-date and whose name
ends in .c.  This is suboptimal, because it will fail to pick up any
changes ever made to CRULE_{NO,}OPT in Makefile.common.
I am kind of assuming that we don't have {variable:Mpattern} on all
the makes that we support ... is there a way to modify CRULE_{NO,}OPT
in-place to effect this transformation?
[21:56:07] <shadow@gmail.com/barnowl754E0B64> can't immediately think of one. hm.
[22:17:25] <kaduk@mit.edu/barnowl> Though, having let that run through for a while, I now get:
make: don't know how to make .o. Stop

Time for sleep before further investigation.
[23:52:49] --- deason/gmail has left