svn commit: r185499 - head

[Bjoern A. Zeeb]

On Mon, 1 Dec 2008, Poul-Henning Kamp wrote:

> In message <863ah8qarb.fsf at ds4.des.no>, =?utf-8?Q?Dag-Erling_Sm=C3=B8rgrav?= wr
> ites:
>> Alfred Perlstein <alfred at freebsd.org> writes:
>>> If you strongly believe that a target should return "true" on failure
>>> and/or output "i worked" on failure
>>
>> I don't, and I don't understand why you think I do.  I have stated very
>> clearly that I don't like the universe target.
>
> I don't mind the universe target failing with the aggregate status code
> at the end, I have said that repeatedly.
>
> But I don't want it to stop working until it has tried compiling
> as much as possible.

so in case anyone with enough make foo can tell me why the initial
echo "0"
is not always (especially on a 2nd and further make universe run)
executed, here's the patch for this:
 	http://people.freebsd.org/~bz/make-universe.diff
I especially had problems when I did make universe; rm _.*; make
universe.

When it comes to giving the tail of the build errors, I think it's
really not usefull for several reasons:
1) if you have lots of build failures you lose the summary and the
    overview the next morning you resume your screen.
2) is it really make -jN clean? What does make do if there are less
    less jobs than told to do parallel builds for a universe? Once it
    starts to subdivide the jobs your build error is somewhere in your
    log but not necessarily at the end.

If you want email notifications etc. go to projcvs and checkout the
real tinderbox instead of overloading our base /Makefile.

/bz

-- 
Bjoern A. Zeeb              Stop bit received. Insert coin for new game.