[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gnat-3.2 transition plan



Matthew Wilcox <willy@debian.org> writes:

> I confess to having made an omission when drawing up the gcc-3.2
> transition plan.  I concentrated on C++ and completely ignored other
> languages.  I do not know what (if anything) nees to be done for Java
> or Fortran.  However, I've done a small amount of investigation on
> what needs to be done for Ada.  Comments and corrections welcomed.

You can ignore Ada for now.  GNAT 5.0 is not a viable replacement for
GNAT 3.14p (or 3.15p).

There is no ASIS, and the compiler is less stable.  For example,
GtkAda (and some other packages) only build after tweaking.

> First problem: arch list.  gnat-3.2 is available on everything except
> arm & m68k.  Rumour has it gnat-3.3 will be available on m68k too.

To be honest, I pretty much doubt that these ports work for anything
except programs which only use a subset of the language (like the GNAT
compiler itself).

> So I propose arch: any for these packages and they will simply fail on
> arm & m68k for the moment.  This seems more sensible than asking package
> maintainers to track which architectures have managed to get gnat working.

Again, just because GNAT is able to compile itself on that
architecture, this doesn't prove it's actually working.

> What should these look like?  I'm tempted to say:
>
>   Build-Depends: gnat

Are you sure you can build GNAT 5.0 in GCC 3.2 using GNAT 3.15p?
There are reports that this doesn't work.

> and
>   Depends: ${shlibs:Depends}
>
> For this to work, I think we need a libgnat3.15a.shlibs file.

Please drop shared library support altogether.  It is currently not
worth the trouble.  GNAT ABIs change from version to version, and the
run-time library can be built only with the corresponding version of
the compiler.  This means that a lot of work is required each time a
new GNAT version is released.

> Anything further needs to be done?

IMHO, the transition to GNAT 5.0 is a non-issue at the moment.  You
can't expect people to work with that compiler in the version which
available from the FSF right now.  And there is no indication
whatsoever that the critical fixes will be applied before sarge hits
the street.



Reply to: