Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!decwrl!sgi!brendan@illyria.wpd.sgi.com
From: brendan@illyria.wpd.sgi.com (Brendan Eich)
Newsgroups: comp.sys.sgi
Subject: Re: Weird command flags
Message-ID: <52669@sgi.sgi.com>
Date: 4 Mar 90 23:40:37 GMT
References: <9003022216.AA27699@blumiris.chem.umr.edu>
Sender: brendan@illyria.wpd.sgi.com
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 41

In article <9003022216.AA27699@blumiris.chem.umr.edu>, bobf@BLUMIRIS.CHEM.UMR.EDU ("Robert B. Funchess") writes:
> A recent posting mentions that cc -s works but doesn't seem to be documented.

cc(1) begins its option documentation by noting "The following options are
interpreted by cc.  See ld(1) for load-time options."  And ld(1) documents
the -s option.

> We are running 3.1 rev D -- this may be fixed in 3.2 but I don't know -- in
> which the man page for install says that -c means "put the file here, iff it
> doesn't already exist".  HOWEVER.... if I try to do this, install gives me
> its built in "dummy, this is the proper syntax" line, which DOESN'T mention
> a -c flag.

The -c option along with -i and -n were dropped from install in 3.1, but the
install(1) man page was not updated.  3.2 includes a man page for /etc/install
that reflects reality.

> Incidentally, the Irix install seems to be -- "enhanced" -- to the point of 
> incompatibility with a lot of Makefiles.  Then again, the Irix make seems to
> be "enhanced" to the point of incompatibility with a lot of Makefiles.

Could you name some of the make enhancements that are incompatible?  SGI
inherited some seldom-used features from MIPS (IPATH, DPATH), and we have
added our own often-used features lately.  But the features I'm aware of
should not create incompatibilities, unless they preempt macro names (as
IPATH and DPATH do).

> Back to the point:  there is an error in either the manual or the install 
> command.  Since I am trying to build TeX, there are one h*** of a lot of 
> Makefiles to wade thru and change every instance of "install" to be Irix-
> compatible.

Is it possible that TeX's makefiles are using the BSD install command, which
has a -c (copy) option?  SGI dropped the SVR3 install -c option because it
was confusingly useless.  When installing from a makefile, the target file's
existence does not mean it's the same as the (usually new) source file being
installed.  AT&T install -c would tend to install your TeX files at most once
in a given tree.  That's why I'm wondering whether TeX's makefiles are not in
fact invoking BSD install -c.

Brendan