Path: gmdzi!unido!mcsun!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!
JANUS.BERKELEY.EDU!talvola
From: talv...@JANUS.BERKELEY.EDU (Erik Talvola)
Newsgroups: gnu.gcc
Subject: Letter about GCC in SunExpert
Message-ID: <9010021857.AA02734@janus.Berkeley.EDU>
Date: 2 Oct 90 18:57:13 GMT
Sender: dae...@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 98
Posted: Tue Oct 2 19:57:13 1990
Here is a letter to the editor which appeared in the September 1990
(Vol. 1 Num. 11) issue of SunExpert which discusses the GNU C compiler.
It would be interesting if maybe the Free Software Foundation would like
to send a rebuttal letter in. The address for SunExpert is:
1330 Beacon Street
Brookline, MA 02146
Email: dpr...@expert.com
-----------
Dear Editor:
Thank you for raising the issue of the unbundling of the Sun C compiler
(SunExpert, June, p. 8). We understand the importance of this issue to the
developer community and we realize that this is a departure from the
traditional bundled UNIX philosophy. And yes, the times they may be a-
changing. A couple of years ago, you could pretty much guarantee that every
Sun workstation sold was going to be a developer tool. However with the
current hardware releases, such as the SLC, many workstations are being sold
that won't be used for software development. This evolution of the hardware
line allows for a change in the software line. We have chosen to unbundle
the C compiler to allow us to invest more resources into this important
developer tool that might not be needed by all of our current, or future,
customers.
It sounds like you agree that the unbundling of the compiler makes sense
from a value-added point of view. The majority of code developed on Sun
workstations is written in C andn the compiler is one of the essential links
in producing top-quality code. By unbundling the compiler we can offer the
developer more frequent updates, substantial improvement to the code
generator, and most importantly, increased responsiveness to customer
requested enhancements.
The unbundling of the C compiler is a recognition by Sun of the importance
of the developer community and an attempt to provide the best tool possible,
although that may mean charging for it. You mentioned the list price of
$2,000, however we ran a special promotion from May through September during
which the compiler was $500. In addition, volume pricing and educational
discounts can significantly reduce this list price. This price also reflects
the inclusion of the Source Browser development tool which is only offered
with the unbundled compiler. In the future we would like to include
additional tools with the compiler product.
You also raise the question of customers drifting to other compilers.
Specifically you mentioned Saber-C and GNU C. Since these are interesting
examples, I will address both.
Saber-C is not an alternative to the Sun C compiler, rather it is a
compliment. Saber-C contains an integrated C interpreter which is essential
to their debugging environment but does not replace a high quality compiler
such as Sun C. In fact, by the time you read this, Saber and Sun will have
announced a copromotion of Saber-C 3.0 and Sun C. We encourage the use of
Saber-C to those developers that need its functionality and we look forward
to the continued support of Saber-C. However, GNU C is a true competitor to
Sun C. It has a number of strong features including fast compilation, a
SPARC back-end, and a rock-bottom price. However, Sun C offers four
distinct advantages over GNU C:
1) More optimizations. Sun C performs 21 global and local optimizations
compared to 11 for GNU C. Sun C does all optimizations that GNU C does plus
important techniques such as loop unrolling, alias optimization, and
instruction scheduling. This letter is not the forum for benchmarks, but I
would be happy to supply your readers with benchmarks from Sun and from our
customers who are developing real systems.
2) Better SPARC code generation. Our team of code generation engineers
work closely with our hardware designers to improve and fine-tune the SPARC-
targeted code. This allows us to release compilers tailored for hardware
enhancements at the time of the hardware release. This provide a competitive
edge over developers using other compilers.
3) Legal issues. In the June 1990 issue of the GNU Bulletin, Richard
Stallman discusses the fact that the GNU General Public License states that
if a developer links GNU libraries with their source code they are then
legally required to distribute all source code with their applications.
This puts a dent into proprietary software development. Sun C has no such
restriction.
4) Support. GNU is a completely unsupported product. GNU used to release
the compiler on a tape labeled "Beta Test" and have recently upgraded to the
"Pre-Release" label. On the other hand, Sun C is a supported product that
includes a full suite of hardcopy documentation and a hotline number for
telephone support, as well as an upgrade and patch release program.
Software development is hard enough as it is without having to figure out
for yourself if the bug is in the code or the compiler.
In conclusion, there is a Zen saying, "When patterns are broken, new
worlds can emerge." With the evolution of our hardware and software lines,
and the commitment to provide the best tools possible, I think that the Sun
community will continue to grow and prosper.
David Reim
C Compiler Product Manager
Sun Microsystems Inc.
-----------
Path: gmdzi!unido!mcsun!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!AI.MIT.EDU!rms
From: r...@AI.MIT.EDU (Richard Stallman)
Newsgroups: gnu.gcc
Subject: letter about GCC in Sun Expert
Message-ID: <9010080203.AA27665@pogo>
Date: 8 Oct 90 02:03:05 GMT
Sender: dae...@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 5
Posted: Mon Oct 8 03:03:05 1990
Rather than having the FSF or any official spokesperson write a
rebuttal, I think it would be much better for the users of GCC to
write what they think of GCC and the claims made about it by Sun.
Rather than rebut one partisan statement with another, let's have
a statement that Sun can't accuse of being partisan.
Path: gmdzi!unido!mcsun!uunet!tut.cis.ohio-state.edu!AI.MIT.EDU!rms
From: r...@AI.MIT.EDU (Richard Stallman)
Newsgroups: gnu.gcc.announce
Subject: Sun Expert
Message-ID: <9010141917.AA07566@pogo>
Date: 14 Oct 90 19:17:09 GMT
Sender: dae...@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 15
Approved: info-...@prep.ai.mit.edu
Posted: Sun Oct 14 20:17:09 1990
Many users have posted messages on the net supporting GCC against the
calumnies published in Sun Expert.
However, posting on the net reaches the wrong set of people. The
people who read net news about GCC already know enough to form their
own opinions. The people who might actually be misled are those who
read Sun Expert and are isolated from other users--the people who will
not see your messages.
So, if you were one of the people who has sent such a message--or if
you are thinking of sending one--please put your message in a letter
and send it to the editor of Sun Expert. Ask them to print it.
It is precisely the opinions held by the users of GCC that carry the
most weight with other users.
Path: gmdzi!unido!mcsun!uunet!tut.cis.ohio-state.edu!AI.MIT.EDU!rms
From: r...@AI.MIT.EDU (Richard Stallman)
Newsgroups: gnu.gcc.announce
Subject: communicating with Sun Expert
Message-ID: <9010170219.AA19535@pogo>
Date: 17 Oct 90 02:19:18 GMT
Sender: dae...@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 7
Approved: info-...@prep.ai.mit.edu
Posted: Wed Oct 17 03:19:18 1990
Send letters to:
Letters to the Editor
SunExpert Magazine
1330 Beacon St.
Brookline, MA 02146
USA
Path: gmdzi!unido!mcsun!uunet!tut.cis.ohio-state.edu!WORLD.STD.COM!bzs
From: b...@WORLD.STD.COM (Barry Shein)
Newsgroups: gnu.gcc.announce
Subject: Re: Letter about GCC in SunExpert
Message-ID: <BZS.90Oct17212627@world.std.com>
Date: 18 Oct 90 01:26:27 GMT
References: <9010021857.AA02734@janus.Berkeley.EDU>
Sender: dae...@tut.cis.ohio-state.edu
Distribution: gnu
Organization: The World
Lines: 38
Approved: info-...@prep.ai.mit.edu
Posted: Thu Oct 18 02:26:27 1990
As Technical Editor for Sun/Expert magazine allow me make a few
comments.
First, the letter from Mr. Reim was in reaction to an editorial we ran
which basically leaned towards GCC and wondered aloud about the sense
of unbundling a product and raising its price when the competition is
free?
So, to a great extent, that letter was published to give the "other
side" an opportunity to respond. Being as it came from someone
directly involved in the issue who was speaking, I assume, for Sun, it
was published without hesitation.
When I read it (before publication) I also expressed the sort of
questions being repeated in this list. But we felt touching it in any
way would appear to be, well, less than forthright. I suspect most of
you can agree with that.
If there is any doubt, I'm glad we have published two views on the
issue now and we are certainly interested in other views. That's what
we're here for, to get ideas into print.
Second, we're interested in all your views. RMS commented that in
order to get to the same audience you have to use the same
publication. This is true.
You can send text intended for inclusion in letters to the editor
either to myself (b...@world.std.com) or Doug Pryor (dpr...@expert.com).
PLEASE clearly mark if you wish to have your e-mail considered for
publication. I suspect some will just want to send me a private note
(which is fine, I'm interested!), so if it's not marked I'll assume
the note was for myself only.
--
-Barry Shein
Software Tool & Die | {xylogics,uunet}!world!bzs | b...@world.std.com
Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD