home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
TeX 1995 July
/
TeX CD-ROM July 1995 (Disc 1)(Walnut Creek)(1995).ISO
/
tex-k
/
tex-k-archive.past
/
1995.02
/
000107_swift@acs.bu.edu_Thu Feb 16 11:20:19 1995.msg
< prev
next >
Wrap
Internet Message Format
|
1995-02-28
|
5KB
Received: from ACS3.BU.EDU by cs.umb.edu with SMTP id AA09903
(5.65c/IDA-1.4.4 for <tex-k@cs.umb.edu>); Thu, 16 Feb 1995 16:24:12 -0500
Received: by acs3.bu.edu (8.6.9/BU_SmartClient-1.0)
id QAA270264; Thu, 16 Feb 1995 16:20:22 -0500
From: swift@acs.bu.edu
Message-Id: <199502162120.QAA270264@acs3.bu.edu>
Date: Thu, 16 Feb 95 16:20:19 -0500
Reply-To: swift@bu.edu
To: bob@gnu.ai.mit.edu, tex-k@cs.umb.edu
Subject: m68k-sony-newsos build
In-Reply-To: (Your message of Thu, 16 Feb 95 13:49:28 EST.)
<9502161849.AA27461@hill.gnu.ai.mit.edu>
I think Bob Chassell's last notice was very well put. I would like to take one
point he made, and reaffirm the arbitrariness in defining a 'vanilla' tex
installation.
> I am not necessarily advocating one or many makefiles. I am advocating a
> top level makefile that enables you to build TeX from scratch. It might also
> let you build other things (or give you clear pointers on how to build
> parts). In a TeX distribution, I expect `make' to build what I need to
> create .dvi files.
You define minimum funcationality as being able to "create .dvi files." The
single program TeX will allow you to create dvi files. But it is very
reasonable to say that one would like to be able to process the example files
like simple.tex, discussed in the TeXbook, which requires the Plain TeX format,
which requires additional library files. A "non-expert" is likely to believe,
quite reasonably, that if he can't process simple.tex, he can't "create dvi
files." To him, "TeX from scratch" includes the plain TeX format. He probably
doesn't want to know about the relationship between Tex the program and the
Plain TeX format, and wants the installation to hide this from him. Then
again, maybe not, maybe he really just wants to build dvi files.
But it doesn't stop there. A .dvi file does most people no good unless more of
the "friends" are around to present it to a human, which requires building
fonts (unless you go for the primitive dvitype or dvi2tty) and a handful of
auxiliary programs to manage them, and building a program to view or to print
the dvi file. Knuth's CMR fonts are standard. Though these parts are not
detailed in the TeXbook, it is assumed in the book that they are there,
installed by an "expert". This perhaps shows its datedness; anyway the TeXbook
does not define a vanilla installation, we are on our own.
Yet other users and installers will be coming to TeX straight from LaTeX or
LaTeX2e, and expect the latex format, the latex fonts, certain common packages,
and so on to be there in a vanilla installation. These people might reasonably
not want to have to learn about the "internal" relationship between TeX and
LaTeX. They consider "TeX from scratch" to be able to process Lamport's or
Shoepf's story.tex (or whatever the 2e little example is called). And so on.
I venture it is impossible to anticipate what a non-expert expects and wants
from a vanilla installation, both because they often don't know what they want,
and different non-experts will want different things, each thinking that what
they want is the perfectly obvious vanilla installation.
It is perhaps less misleading to acknowledge up front to a potential installer
that this is not going to be as simple a process as with other pieces of
softaware: the process will require learning enough details to figure out
what is wanted. No amount of user-friendliness and automation can
anticipate this.
I think tex's amorphousness distinguishes it from other large package-oriented
software like GNU emacs, which despite its huge flexibility and number of
library files still has a structure that places a single binary file in the
center (spread across the autoloaded lisp files, I suppose) and all other
programs as strictly subordinate and unnecessary for vanilla operation, which
in the case of emacs is easier to define: load, maniputlate, and save text
files from a terminal and/or X windows. No fifty million kinds of output
devices for example.
Part of the question to be answered here is, until eventually the distribution
is put into ideal form, to whom should the maintainers devote more of their
finite time, the beginner or the person with some knowledge of tex installation
structure?
This difficulty only bears on one aspect of Bob's discussion, which otherwise
as I say convinced me as sound thinking, e.g., the directory expectations
during building. These expectations, however, assume a source tar.gz file that
is _entirely_ self-sufficent (the point of difficulty discussed above). web2c
isn't self-sufficent; many people want to use what it contains, and adapt it to
an existing library structure. If you don't have an existing library
structure, or want to adopt the new standard one, you get one more tar.gz file,
and plonk it in. It does indeed require some knowledge on the part of the
installer about what they want and need; but in my opinion as I said above,
this is less misleading: I think a greater number of people will be more
greatly confused by an installation that appears and installs like a "vanilla"
installation, but turns out to be French Vanilla, that is, someone else's idea
of vanilla that doesn't taste good.
regards,
Matt