ghoti@BOURBAKI.MIT.EDU (01/24/89)
Enclosed is my recent correspondence with Richard Stallman regarding
my suggestions for making GCC source code more accessible to amateurs,
such as myself, who are interested in studying and experimenting with
GCC source code. It includes an outline for the book I wish someone would
write.
Sincerely,
Allan Adler
ghoti@hypatia.mit.edu
===========================================================================
Adler to Stallman
I'm glad that you like the idea of a book. I am also grateful that you are
willing to answer questions. I think that copyleft for the book is a great
idea. That does pose some problems regarding publication and distribution
but I think they can be easily handled. However, I should explain the
constraints I am working under since they are severe:
(1) I have been unemployed for 3 years and am destitute. I live
entirely on charity, which is by no means a mode of survival.
Furthermore, because of my strong ideals regarding education and
employment practices, and because of my commitment to my own
scholarly interests, it is quite unlikely that things can ever
get better for me (although they can easily get worse).
(2) I don't have enough room on the machine I'm using to store a
reasonably current version of GCC. Furthermore, I have no
guarantee of being to keep my account for long enough to
finish the project.
(3) I am not competent to write a book on implementing GCC. I have
only attempted one implementation (namely to a nonexistent
Z80 based system for which I have a half-finished simulator)
and it was unsuccessful because I couldn't figure out how to
debug it.
(4) Although I am quite enthusiastic about the educational potential
of GCC and FSF software in general, my strongest interests lie
in other areas such as pure mathematics, musical composition,
history, etc. In view of (1), this means that if I have to go
down fighting (as seems likely), it will be for such studies,
not for FSF. Furthermore, even if (1) were not a consideration,
I would still give considerable priority to my own scholarship.
(5) Even with your generous offer to answer questions, I think I will
need more help than you can afford to provide. I will therefore
have to get other people involved in the effort, and so far
I haven't been able to find other people interested in helping me.
One way to solve some of these problems would be for me to be part of a grant
proposal. I probably couldn't be PI for one since my Ph.D. is in mathematics,
not computer science. But I could probably be listed as a consultant, where
my mathematical background and my experience as an educator would be relevant.
But someone else would have to ask for the grant. But who ?
As for the contents of the book itself, here is a naive and tentative outline.
In view of (3) above, I am reminded of when, as an experiment, I gave my class
the homework problem of writing a midterm which they thought would be
reasonable. I discovered that the most difficult exams were written by the
students least likely to be able to pass them. And so it is with the outline
which I submit to you below for your comments, composed by the person least
likely to be able to fill in the details.
Sincerely,
Allan Adler
ghoti@hypatia.mit.edu
===============================================================================
Outline, submitted by Allan Adler
I. Overview of this document
II. The machines on which GCC is implemented
A. Prose description of those characteristics of target machines and
their operating systems which are pertinent to the implementation
of GCC.
i. target machine characteristics
ii. comparison of this system of machine characteristics with
other systems of machine specification currently in use
B. Tables of the characteristics of the target machines on which
GCC is implemented.
C. A priori restrictions on the values of characteristics of target
machines, with special reference to those parts of the source code
which entail those restrictions
i. restrictions on individual characteristics
ii. restrictions involving relations between various characteristics
D. The use of GCC as a cross-compiler
i. cross-compilation to existing machines
ii. cross-compilation to non-existent machines
E. Assumptions which are made regarding the target machine
i. assumptions regarding the architecture of the target machine
ii. assumptions regarding the operating system of the target machine
iii. assumptions regarding software resident on the target machine
iv. where in the source code these assumptions are in effect made
III. A systematic approach to writing the files tm.h and xm.h
A. Macros which require special care or skill
B. Examples, with special reference to the underlying reasoning
IV. A systematic approach to writing the machine description file md
A. Minimal viable sets of insn descriptions
B. Examples, with special reference to the underlying reasoning
V. Uses of the file aux-output.c
VI. Specification of what constitutes correct performance of an
implementation of GCC
A. The components of GCC
B. The itinerary of a C program through GCC
C. A system of benchmark programs
D. What your RTL and assembler look should like with these benchmarks
E. Fine tuning for speed and size of compiled code
VII. A systematic procedure for debugging an implementation of GCC
A. Use of GDB
i. the macros in the file .gdbinit for GCC
ii. obstacles to the use of GDB
a. functions implemented as macros can't be called in GDB
B. Debugging without GDB
i. How to turn off the normal operation of GCC and access specific
parts of GCC
C. Debugging tools
D. Debugging the parser
i. how machine macros and insn definitions affect the parser
E. Debugging the recognizer
i. how machine macros and insn definitions affect the recognizer
ditto for other components
VIII. Data types used by GCC, with special reference to restrictions not
specified in the formal definition of the data type
IX. References to the literature for the algorithms used in GCC
===============================================================================
Stallman to Adler
Received: from sugar-bombs.ai.mit.edu by math.mit.edu;
Tue, 24 Jan 89 02:33:38 EST
Received: by sugar-bombs.ai.mit.edu; Tue, 24 Jan 89 02:33:22 EST
Date: Tue, 24 Jan 89 02:33:22 EST
From: rms@wheaties.ai.mit.edu (Richard Stallman)
Message-Id: <8901240733.AA00701@sugar-bombs.ai.mit.edu>
To: ghoti@bourbaki.mit.edu
In-Reply-To: <8901240655.AA28576@hermite.mit.edu> (ghoti@bourbaki.mit.edu)
Based on what you say, I don't think you are the person to write
a book on GCC. But your outline looks reasonable, so I will put it in
PROJECTS in case someone shows up wanting to write the stuff.