Re: GNUmake compared with Watcom make

[Johan Bezem]

Michael Mounteney wrote:
>...
> << I try to avoid building over a network, and objects files are *always*
> generated locally. I[t] simply takes too long. >>
> 
> I then get trouble with timestamps; 

On a windows-build, I have sources and objects on the same machine,
synchronized with my (Solaris-based) homedir using "CopyTo" 2.6
(http://www.ne.jp/asahi/cool/kish/) over Samba, on Solaris I haven't had any
problems storing objects on a local disk, and sources/binaries/libs on an
nfs-based homedir.

> ...also, I sometimes reboot the Pentium III
> laptop into W'2000 to do a clean build, so the object files have to be on a
> filesystem that's visible to both laptops.

I do not quite follow, but I suppose you have multiple config's on one
machine, and both configs need access to the sources. Why do you need access
to the objects?

> << I have to assume that the makefile complexity is identical, otherwise these
> numbers mean nothing.  Go hunting for superflous spawn's and shell's, GNUmake
> can be very unforgiving ;-) >>
> 
> Surely on a null build, there should be *no* spawning or execing ?  I use one
> monster makefile, not a per-directory hierarchy of makefiles.  It seems to me
> that once make (GNU or w) has loaded, all it has to do is check time stamps.

Sure, if you have one monstermake, make will not start a process for every
directory. However, make has some specific features that will possibly spawn
at least some processes, in addition to the shell(s?) neede to execute the
normal commands that _will_ get executed even on a null-make (using '+', for
instance, or variable settings using $(shell), as well as a few others...).
Depends on your makefile.

Ciao,

Johan Bezem
CSK Software AG