Subject: Re: Thoughts on GPL
From: "Jonathan S. Shapiro" <jsshapiro@earthlink.net>
Date: Tue, 24 Feb 1998 20:37:53 -0500

    > It not economically justifiable to invest in
    > competing compilers, nor is it economically supportable to build a
    > marketing and sales organization to sell GCC in large numbers.

   Seems to me that Cygnus is trying to do just that.  Haven't you seen
   their ads in Electronics Week or EDN or whatever tome it is that drops 
   into my mailbox weekly?

I don't recall Cygnus doing this prior to their venture investment.  I
contend that what you are seeing is a transfer payment derived from
their strictly commercial activities, and that the GPL model doesn't
work on its own.

    > Microsoft is still able to sell because the GPL community largely
    > ignores Windows and because Microsoft has monopoly control on the
    > API's and alters them with high frequency.

   "and because"?  You mean "because".

I meant "and because".  There are two distinct reasons:

  1. The GPL community dislikes windows
  2. The monopoly protection lies in the rate of change to the API,
     not in the compiler.

The first *may* be because of the second, but I think a lot of it is
just a visceral dislike of Windows.

    > Meanwhile, Cygnus *has* succeeded in making the compiler industry very
    > difficult to make money in.  Meanwhile, the rate of evolution on GCC
    > has dropped in recent years.

   Which gcc?  The FSF gcc or the gcc you get from Cygnus?  Cygnus has
   been reluctant to fork a version, but finally they had to create egcs, 
   which is evolving, dare I say, rapidly?

Note that EGCS cannot be readily found from the Web site.  Cygnus
clearly does not view the project as important to their core
positioning.  

Also, the very fact that they needed to fork is a circumstantial
indicator that the system isn't working.

    > So who won?  Nobody.  The customers, in the end, have also lost due to
    > lack of competition.

   Arguably.  Arguably they've won because of the money saved by not
   having to buy compilers.  Arguably they've won by having a stable
   compiler with known bugs.  Ever discovered a compiler bug?  It'll ruin
   your whole week -- especially when you call customer support only to
   find out that it's a known but unpublished bug.  Unpublished for
   competitive reasons.

Actually, I've run compiler groups for a living, so I'm more than
familiar with the problems.  All of the issues you identify are real,
but they are all short-term.  In the long term, I contend, evolution
has been seriously hampered.


shap