Date: Wed, 26 Feb 1997 00:19:55 -0800 (PST) From: asami@vader.cs.berkeley.edu (Satoshi Asami) To: adam@veda.is Cc: imp@village.ORG, freebsd-ports@FreeBSD.ORG Subject: Re: make -k oddities Message-ID: <199702260819.AAA04603@silvia.HIP.Berkeley.EDU> In-Reply-To: <199702260818.IAA08085@veda.is> (message from Adam David on Wed, 26 Feb 1997 08:18:18 %2B0000 (GMT))
next in thread | previous in thread | raw e-mail | index | archive | help
* Fundamental command paradigm
*
* user: do this
* unix: (ok)
*
* :) silence is golden :)
*
* Successfully completing the command hardly counts as an error.
* This is similar to rm -f.
I don't think so. It will bitch and whine and say "<target> was not
remade because of errors."
* If make -k returns the command status of the most recent command, there is
* no guarantee that it will have been the failed command. If you would let
* make -k return 1 to signify any failed command or 2 for abnormal condition,
* there is no indication of which command(s) failed, if that matters.
If that matters, you shouldn't be using "make -k". :)
Actually I think your second suggestion is better, to have "make -k"
return 1 for any error. I said "ok, build a and b and c, but if one
of them didn't build, don't stop and do your best with the rest."
make will say "ok" and return 0 only if all of them built
successfully.
Note that even with "-k", make will check the dependency and sequence
of commands to ensure not to continue execution on a broken path. If
"-k" is passed to sub-makes in ${.MAKEFLAGS}, it will try to continue
(and this is the problem that brought up the current discussion).
* Would you also let -command pass an error when make exits? (why, why not?)
That is different. "-" is an explicit "I don't care what this command
returns" in the *Makefile*. make has no business messing with their
return status.
Satoshi
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199702260819.AAA04603>