Path: utzoo!attcan!uunet!lll-winken!ncis.llnl.gov!helios.ee.lbl.gov!pasteur!ucbvax!tut.cis.ohio-state.edu!bloom-beacon!EXPO.LCS.MIT.EDU!jim
From: jim@EXPO.LCS.MIT.EDU (Jim Fulton)
Newsgroups: comp.windows.x
Subject: Re: NON-Stupid Imake question
Message-ID: <8903012343.AA14597@expo.lcs.mit.edu>
Date: 1 Mar 89 23:43:09 GMT
References: <3462@stiatl.UUCP>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: X Consortium, MIT Laboratory for Computer Science
Lines: 31


> I beg to differ. A WELL-done Makefile needs little or no editing.

Even a well done Makefile can't guess at what special options the compiler
needs, what configuration options you've enabled, where the software should get
installed, what special make directives or rules have to be inserted, what
extra libraries need to be linked in, what special commands need to be run,
etc.  Trying to remember all of that junk is a waste of brain cells.

If you have to edit an Imakefile you should report it to the author.


>          Imake is not a panacea - you still have to think, design,
> and implement with knowledge & undersatanding of the issues to get
> Imakefiles to work as advertised.

I don't anybody ever claimed imake was perfect.  Imakefiles are most easily
written by finding one from a similar library or program and tweaking it.

Makefiles are the assembler code of software management.

							Jim


p.s. This might be a good time to mention that contributed software for R4
will *require* Imakefiles.

p.p.s.  People who are interested in imake should get a copy of the proceedings
of the USENIX Software Management Workshop to be held in April.  Assuming
UPS Overnight got there in time, there should be a paper describing how and
why the MIT releases of X use imake.