Jerry van Dijk has long involvement in past ports of GNAT (Ada) to win32 by
ACT. I have asked him if he could help with mingw and he has agreed. I
propose that we invite Jerry to join mingw-dvlpr with specific role in GNAT
--- Jerry van Dijk <jvandyk@...> wrote: > From: "Jerry van Dijk"
> Date: Mon, 10 Jun 2002 10:11:52 +0200
> To: Danny Smith <danny_r_smith_2001@...>
> Subject: Re: [Mingw-users] Re: Ada compiler not configured correctly
> Danny Smith writes:
> > That is no handicap. Would you like to be the current maintainer for GNAT
> > mingw (or at least be involved as developer)? If so, I will forward this
> > mingw-devlpr list.
> I had to think about that, like everyones my spare time is limited. But it
> is also important to have a good GNAT implementation on Windows. So I can
> try to help out.
> I'll have to start with setting up a build environment (do co's from CVS)
> to get debug versions of the compiler. Note that I am not using MSYS, but
> have my own environment running on Win/Me. Based on the MKS toolkit, with
> modified Win32 utils added. I have been building 3.x since its pre-release
> so I know it works. Also I only have modem connection (and pay by second
> telephone), so I am not constantly on-line.
> Things that need to be done (IMHO, of course):
> 1. Check that the compiler actually works on Windows (and not just
> NT/W2K/XPro). Encourage Windows users to try the beta and report
> bugs reaching out on the comp.lang.ada newsgroup and the various
> ada mailing lists.
> 2. Setup a CVS repository for Ada regression testing, to be filled from the
> bug reports, with some scripts to run them. I'm not sure dejagnu will
> be useful for this, as external reports will vary wildly (talking from
> experience here :-)
> 3. Laurent Guerby is trying to incorporate the ACATS (Ada Conformity
> Assessment Test Suite) into dejagnu. I don't know the current status
> of this. Meanwhile I have my own scripts to run this on Windows. It
> will help to flesh out bugs and show that the actual language is
> implemented :-). Note the you cannot get a clean ACATS run on Windows
> due to win32 limitations in the Tasking. Suggestion: run the last ACT
> public windows release (3.14p) through ACATS and aim for identical
> results. Although some problems will be 3.x related and not mingw based.
> 4. GNAT has more and more extensions, both in the compiler (special
> attributes and pragma's) and in tools (e.g. GNAT project files).
> Tests have to be gotton from ACT (Geert Bosch) or devised to check
> them on Windows.
> 5. Add tools like gnatreg in a suitable fashion and add Ada installation
> instructions. Which should at least also point to the additional tools
> most people use: ADAGide, GvD, GtkAda, GNATCOM/GWindows, and such.
> 6. Get the authors of these tools and important libraries (Booch components,
> Ada-Sockets, GLADE, Ada Web Server, CORBA, etc). to test them with the
> mingw compiler.
> 7. Try to get the gdb patches from ACT to add Ada-Awareness (including
> tasking and exceptions) to gdb. This might mean going back to an
> older version of gdb for Ada users.
> 8. Create a new Win32 API binding for Ada from the w32api distribution.
> This will remove the MS dependency, make it easier to write multiple
> language programs, will be more up to date then the current old
> Win32Ada binding and can, finally, also be more efficient as it
> can use the special GNAT extensions developped for this purpose.
> Well, that's enough work for the coming year :-)
> Priority of course has testing and bug fixing, followed by the Win32 binding
> (so that tool and library authors can build with the mingw compiler).
> I guess I will also have to get back on the gcc list...
> -- Jerry van Dijk | email: jvandyk@...
> -- Leiden, Holland | web: users.ncrvnet.nl/gmvdijk
http://www.sold.com.au - SOLD.com.au
- Find yourself a bargain!