Subject: Q -- how clean is realclean?
From: Paul Rohr (paul@abisource.com)
Date: Fri Jun 09 2000 - 18:46:06 CDT
[ NOTE: The following question is *not* hypothetical. I updated my build
today to try out all the cool new libole stuff and even the realclean build
failed on a getopt problem. However, it worked just fine after I triggered
the "hidden" realclean target. A similar fix may be all it takes to get the
BeOS PPC build to finish. ]
A while back, I noticed that our realclean target (abi/src/Makefile) doesn't
actually force our peer libraries to rerun the configure scripts. Since the
required peer libraries differ on a per-platform basis, I went ahead and
added another realclean target at the appropriate level:
abi/src/config/require/win/Makefile
The net effect is that there are currently two levels of cleanliness you can
currently achieve:
1. cd /work/abi/; make realclean
---------------------------------
This blows away all the .o[bj] files needed to build AbiWord, but leaves all
the configure-generated header files intact. In most cases, this is just
fine.
Indeed, in many cases, Sam's less-intrusive abiclean target will do the job,
since it also keeps the peer .o[bj]s intact.
2. cd /work/abi/config/require/win/ ; make realclean
-----------------------------------------------------
However, in the ultra-rare cases where you need to regenerate the peer
headers, you need to know to do this second step. (And most of us don't.)
the question
------------
Should we fix the top-level realclean target to also trigger the lower-level
one? This shouldn't be hard for your average makefile hacker, but I'm not
sure whether people will find it desirable or not.
Paul
This archive was generated by hypermail 2b25 : Fri Jun 09 2000 - 18:40:30 CDT