Re: GNAT on Win 95

[Bob Crispen <revbob@EIGHT-BALL.HV.BOEING.COM>]

Robert Dewar <dewar@CS.NYU.EDU> sez:

>"Ditto for Cygnus and LabTek -- here we've got two groups of people
>(saintly people) writing a gcc for Windows 95 and NT.  Both of them
>compile C.  One of them compiles C++.  One of them compiles Ada 95.
>It's a virtual certainty that the same problems are being solved twice.
>It's also a virtual certainty that every day the two common cores of
>the compilers move another millimeter farther apart, and the chances
>of forever having to have two compilers on my machine increase another"
>
>What on earth are you talking about. LabTek is not writing a "gcc for
>Windows 95"! The port of GNAT for NT uses the current release of gcc
>from FSF with a few extra patches which represent changes present
>in the current development version of FSF gcc which are necessary
>for GNAT>

Oh, very well.  Cygnus and Labtek are separately developing
compilers in the gcc family.  Is that precise enough?  You clearly
knew from your other remarks that I was not alleging that these were
clean-room efforts.

>(the maintanence of the FSF version of gcc, which is what
>Labtek -- and all other versions of GNAT -- uses, is under the
>control and supervision of Richard Kenner of ACT and NYU)
>
>The Cygnus development of gcc is kept syncrhonized with the FSF
>development, and Cygnus fixes and improvements are incorporated
>into the FSF release.

Excellent news.  Perhaps I'm not the only one to whom it is news.
Would you be kind enough to mention where, apart from here and now,
you've actually said that with respect to those two compilers?

>"One of them compiles C++"
>
>more massive confusion. gcc is a driver program that can call cc1, gnat1,
>or cc1plus etc, which are the actual compilers for the different languages.
>Each of these compilers incorporates the backend code generator etc that
>is a fundamental part of gcc.

The release notes for the last version of gnat that I installed on
my Sun mentioned what was required to use the same front end for C++,
which is why I knew it was possible on some platforms.  But since
the gnat release notes for Win95 not only did not mention that but
said that C++ was not supported, my statement becomes a *little*
easier to justify.

>If you get the latest version of the gcc driver from either FSF or Cygnus
>then they should be, barring minor revision shifts, identical, and either
>of them will be happy to load the C++ compiler or the GNAT compiler.
[moved]
>Bob, you really ought to take the effort to find more about what
>is going on and how gcc works before posting a message which is
>so confused.

Well, that "barring minor revision shifts" tells me that you, who
of all people in the world should know, don't know either whether the
current releases of the Cygnus and LabTek compilers will permit that!
That's really too much: scolding a compiler user for not knowing what
you, a compiler developer, do not know.

Bob Crispen
revbob@eight-ball.hv.boeing.com
Speaking for myself, not my company