From: faith@cs.unc.edu (Rik Faith) Subject: Re: Compiling Linux with make 3.63 (ANSWER) and faster gzip 0.8.2 Date: 28 Jan 1993 12:52:23 GMT
In article <1993Jan27.202130.5896@adv.win.tue.nl>, devet@adv.win.tue.nl (Arjan de Vet) writes:
|> In article <1993Jan24.223004.8078@tc.cornell.edu> faith@cs.unc.edu (Rik
|> Faith) writes:
|>
|> >ANNOUNCE: BINARIES UPLOADED!
|>
|> >gzip081A.Notes
|> >gzip081A.taz
|>
|> There's already a newer version 0.8.2 which detects correctly that Linux is
|> a 386 system and uses some assembly code for string matching. This gives a
|> speed-up of 30-40%. Compiles and configures without any problem, try it!
You're a few days behind. This was posted before 0.8.2 was out.
The changes between 0.8.1 and 0.8.2 were due to fast work by Stephen
Tweedie <sct@dcs.edinburgh.ac.uk> and Jean-loup Gailly (jloup@chorus.fr).
|> >mk363A.Notes
|> >mk363A.taz
|> >
|> > This is GNU make 3.63, with man and info pages. WARNING: Variables are
|> > no longer automatically exported to the environment when a sub-make is
|> > started. This radically changes behavior of sub-makes! The Linux
|> > kernel will no longer compile without minor changes (which will
|> > probably be in 0.99.5)! USE WITH CAUTION! mk362A.taz does not have
|> > this behavior, and will not be deleted for some time.
|>
|> The fix is very simple:
|>
|> Add:
|>
|> .EXPORT_ALL_VARIABLES:
|>
|> as first line to /usr/src/linux/Makefile.
And in some cases, the builtin rules for submakes will become corrupt.
This is evident if you compile the SCSI code, and is the fault of a *BUG*
in make, which is due to the changes that I warn about. Patches should be
out soon.
-- Rik Faith: faith@cs.unc.edu What I tell you three times is true. --Dodgson