Re: [patch] parallel make bug with failing commands

[Michael Matz]

Hi Paul,

On Sun, 7 Aug 2005 address@hidden wrote:

> Thanks for a most excellent bug report!  I was able to easily follow the
> path you described and see the bug.  I used essentially your fix (with a
> few minor syntactic tweaks).

Nice.

> FYI, following your notes I was able to produce a makefile which is
> guaranteed to show the bug every time, regardless of # of CPUs or other
> factors.  The trick is to use sleep to ensure that the various targets
> complete in the proper order to display the bug; see below.

Ah, right, good idea.  Btw. funnily I can't reproduce the effect of the
deferred output on one of the machines even with your makefile.  The child
is left running, but the echo "ok done" is missing, probably because the
parent make is done, and hence can't execute further commands of the list.  
So to make it even more reproducable one should simply make the sleep and 
the "echo" be one command, then it's guaranteed to be form disturbing 
output in the prompt :)  Ala:

ok:
        @sleep 4; echo ok done


Ciao,
Michael.