*BSD News Article 49799


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news.sprintlink.net!in1.uu.net!solaris.cc.vt.edu!csugrad.cs.vt.edu!csugrad.cs.vt.edu!not-for-mail
From: ceharris@csugrad.cs.vt.edu (Carl Harris)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: pmake + FreeBSD
Date: 24 Aug 1995 20:35:55 -0400
Organization: Virginia Tech Computer Science
Lines: 48
Message-ID: <41j5tb$nli@csugrad.cs.vt.edu>
References: <418nhj$vq2@csugrad.cs.vt.edu> <41cjts$1ol@bonnie.tcd-dresden.de> <41fh2r$gt2@csugrad.cs.vt.edu> <41hhg4$9g0@bonnie.tcd-dresden.de>
Reply-To: ceharris@vt.edu
NNTP-Posting-Host: csugrad.cs.vt.edu
X-Newsreader: TIN [version 1.2 PL2]

J Wunsch (j@bonnie.heep.sax.de) wrote:
: >It's not the Berkeley-isms in the Makefiles, it's various FreeBSD-isms
: >in the build.  Things like sys_errlist, sys_siglist, etc being defined

: It's been there for long time (and logically belongs there):

I'm not arguing that it shouldn't be there.  I actually agree that it
should be there.  My point was more that it is a royal pain to go back
into someone's code and rip out places where the program shouldn't be
defining such things.

: And 'ar' forcing the use of - before the command,

: Ain't this just Posix?  All programs do it this way.  If historical
: version didn't, they are broken^H^H^H^H^Hinconsistent.

Agreed.  Again, I'm not faulting FreeBSD.  Or Posix.  Or ar.  Just another
annoying little thing to fix.


: My original question rather was: what would pmake buy you that the
: current (b)make does not do?

Unless I'm really missing something in the (b)make man page, it has
none of pmake's parallel make facilities (i.e. there is no customs
daemon, there is no reginfo utility to check to see what clients are
willing to handle parts of the make, etc).  I have a lab with 20
P75-based FreeBSD machines.  With pmake, I can use the whole lab as a
processor array for building several large projects using pmake's
RPC-based parallel make facility.  Note that pmake does not use 
Sun/ONC RPC.  Rather it provides it's own RPC module.  

I already use pmake on a cluster of 10 DECstations.  I have another 
cluster of four 175MHz DEC Alphas to do the same thing for that 
architecture.  I would just like to get it working on FreeBSD, too, 
since I have plenty of Pentium cycles to burn, and lots of projects 
to build for that processor.

If you haven't used pmake for parallel make, and you have access to 
enough processors to make it worth your while, I highly recommend pmake
over all of the other parallel makes I've seen for U*ix.  I should have
it working on FreeBSD Real Soon Now.

--
Carl Harris
EXECUTIVE Scapegoat (and Systems Engineer)
Department of Computer Science, Virginia Tech
ceharris@cs.vt.edu