This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Porting GCC to a new OS


> Date: Fri, 12 Nov 1999 15:20:22 -0600
> From: David Starner <dvdeug@x8b4e53cd.dhcp.okstate.edu>

> One of the first things they need is an Ada compiler. There has been
> discussions about porting GCC, but there have been worries that it
> will be difficult considering that the OS will not be a Unix-style
> system.

gcc is the most ported compiler on the face of this earth.  No
existing compiler can be easier, generally speaking.

You can write your own compiler, Ada frontend, optimizer, machine
port, if you want.  But in the process you _will_ have to replicate
100 man years or more to get near where you would start with if you
use gcc.

> How difficult will it be to port GCC cleanly

First, there are two issues.  Before you have a system, you will need
to compile code _for_ the system.  This means you will need a cross
compiler and a host system.  If you choose Linux, or Solaris or Win95
even, you will find that gcc `just' works.  You'll have to port what
runtime libraries you want for your new system.  This shouldn't be
that hard.

Long after you have the cross compiler working, you can then see about
hosting the compiler on that system.  gcc doesn't need that much in
the way of OS resources (open/read/write type things with an odd
fork/exec), my guess is that you could easily do with with less than
100 man years of work, easily making it worth while.

Or, you could waste your time, and reinvent the world.

By the way, you should go scout out the GNU Ada folks, they don't hang
out here much.  They might be able to say more about what the Ada
runtime might need or require.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]