Mail Archives: djgpp/1997/10/05/09:18:13
On Mon, 29 Sep 1997 08:02:44 GMT, Eli Zaretskii <eliz AT is DOT elta DOT co DOT il> wrote:
>Anyway, it only takes a change as simple as removing the extra blank
>between "target1" and the colon, and your Makefile works:
Ah, so THAT'S it. I ran into similar problems with MIDAS Digital Audio
System when we upgraded to Make 3.76, and couldn't figure those strange
error messages out. We have about 35kb of makefiles just to get all
settings and macros right, plus the usual dependency and target
information (25kb more or so).
>And btw, why did you have to use this weird way of choosing between
>two targets? I think the following is much better:
I personally used this to print error messages from the makefiles if the
user has set incorrect or conflicting settings. For example, if no host
system has been set, and the makefiles haven't been able to figure them
out themselves, the target
ifndef _SYS
error:
echo Invalid system $(SYSTEM)!
endif
makes sure that the user has at least some chance of hearing about this.
You can probably guess where the extra space was in my system :)
These "error" targets are before the normal default target, so they get
built if something goes wrong. Ideas for a better way to handle this are
welcome.
BTW, have I mentioned lately that I hate GNU Make? A lot. With passion.
Visual C runtime library wins though - after some problems with my
computer ending the daylight savings time a month too early, VCs stat()
and time() now return times with a one hour difference for a file that has
just been modified. When all files seem to be modified one hour in the
future, you can probably guess how well make works... Plus for some
interesting reason VCs time() returns values 4h bigger than Watcom C or
GCC. Thank god for the Cygnus Win32 toolset.
Petteri - "make BUILD=retail TARGET=dos cleantarget clean lib"
- Raw text -