This is the mail archive of the
mailing list for the GCC project.
Re: obseleting VMS ports?
- From: Nathanael Nerode <neroden at doctormoo dot dyndns dot org>
- To: gcc at gcc dot gnu dot org
- Date: Mon, 27 May 2002 15:06:33 -0400
- Subject: Re: obseleting VMS ports?
- References: <20020527181326.16E42F28CC@nile.gnat.com>
On Mon, May 27, 2002 at 02:13:26PM -0400, Robert Dewar wrote:
> <<I'd like to suggest that all the *-*-vms* targets be obseleted, unless
> people can provide accurate reports that they have 3.x-series gcc
> running and working for a particular target, and that they're
> actually using it.
> We bootstrap GCC 3.1 all the time on OpenVMS. There are some problems with
> the GNAT port, but it's by no means the worst one. So please do not remove
> these targets, or more accurately, if these are removed, then we would
> have to fork completely since we depend on the vms support.
OK, sounds like alpha-*-*vms* has a reason to live.
The existing documentation describes a complicated scheme for
transforming unix-style paths into VMS-style paths. Is this still used?
Is it needed? Does it work? Where is the code located and who maintains
it (if anyone)?
It says that GCC with a VMS target *requires* GAS and *requires* the VMS
linker (not GNU ld). Is this accurate? Any other weird caveats? Do
both the VMS debugger and GDB work, or does only one?
It looks from the code in the config directory like you can't
cross-compile from anything else to VMS. Or maybe you just can't
cross-*link*. If you can't cross-compile, that should be documented.
If you just can't cross-link, perhaps a few notes on how to deal would
It's not clear what non-gcc C compilers under VMS can be used to compile
GCC, or whether there are any caveats when doing so.
It's unclear whether building gcc/gdb/binutils in the same directory is
a good idea, or a bad idea. (on most machines it's a good idea, but...)
It looks like GNU make is mandatory, but it's neither clear nor