On Sat, 14 Aug 1999, Shigeki Uno wrote:

> It took "82" hours (3days and 10 hours). Can you believe that ?
> I live in Japan and can go to The States and return here during
> "82" hours. SIGH-SIGH-SIGH-SIGH-SIGH
> 
> I did make build every tarballs. But If I've already installed binaries
> into /bin ,/usr/bin ,/sbin and  /usr/sbin, I don't have to install them, 
> I think. 
> Am I wrong ? Are there any other ways to save time to make build ?

I'm not following you. "make release" does a "make build", then makes
the tarballs. "make build" compiles and installs the new binaries in
place. If it's just for that one machine, "make build" is all you need.

> LC520 (68030/25 with fpu) 36mb ram/160mb intenal & 1gb external HDD

On the 840AV, it took ~24 hours to build the exportable release, so
your result isn't surprising. If you're following -current frequently,
you can shave a few hours by skipping the man pages, as "make MKMAN=no
build". Also, when you upgrade the compiler, as with 1.4.x -> 1.5x,
you really should "make build" ***twice***, but note that you never
need to build man pages for the first "bootstrap" build.

If you leave your tree populated, you may be able to save quite a bit
of time with "make UPDATE=1 BUILD=1 build", but _be_ _careful_. If
anything is moved your build may break, and you have no right to
complain until you've done "make clean" and tried again. If you're
going to use UPDATE, "sup" is out, because it backdates the files and
messes things all up. UPDATE should work better with "anoncvs".

Best to ask for details on current-users. Except for the amount of
time involved, the build works about the same on all ports. If there
is something broken just on mac68k, be sure to file pr's and whine
loudly. Otherwise, what's the point of running -current?
;-)