[talk] possible meeting topic: FreeBSD desktop stuff with a side of DJ-BSD redux

[James K. Lowden]

On Mon, 3 Mar 2025 17:36:11 -0500
Charlie Li <vishwin at freebsd.org> wrote:

> is interest. There is now a FreeBSD Laptop Desktop Working Group
> (LDWG) [0] that meets monthly (on Jitsi), with some Foundation
> involvement, 

Can I talk to you about gcc?  FSVO $talk and  $you? 

As perhaps some here assembled may know, I've been working on adding
COBOL to gcc for the last, oh, 4 years.  The idea was adopted by the
GCC steering committee a couple years back, and now we're in the
midst of having our patches applied upstream, hopefully for inclusion
in gcc-15, due for release next year.  Yes, we're already talking about
mid-2026.  

For expediency, we've been developing in Ubuntu.  It's ~everywhere; for
example I use Ubuntu Multipass on macOS to install Linux virtual
machines on my laptop.  I don't guess LDWG cares about that kind of
environment, but it's one reason (of many) that Linux is where we
started.  

I was contacted upon a time about patches for FreeBSD, but that
conversation petered out.  I would still like to make sure the compiler
and runtime library build and install happily on FreeBSD (well, *BSD).
I'm happy to work with anyone who shares that goal.  

Some may ask: Why?  Why COBOL?  If I may be permitted....

The answer is simply this: 100 billlion lines or more.  There is a lot
of COBOL out there you've never seen.  There is a (measured) 95%
probability your last ATM transaction went through a COBOL application
written in the 1980s, if not earlier.  The books-and-records
applications written for the mainframe Back In The Day still run,
because computing interest hasn't changed.  Actually: *can't* change,
because no one wants their interest calculated differently.  

Can't change is why those applications still run on the mainframe,
using emulation provided by IBM for machines they haven't manufactured
in decades.  Rest assured, it works.  It has to. it is also Not Cheap.  

Mayhap you, dear reader, work for such an organization.  If all goes
according to plan, there will be two open-source compilers for COBOL:
GnuCOBOL, which converts COBOL to C, and uses gcc to compile that, and
gcc-cobol, which is just gcc.  That means two ways to "lift and
shift": compile the unaltered COBOL source code for any gcc target, and
run it without paying IBM's totally reasonable mainframe licensing
fees.  

[pitch] 
I can help, but I don't have to.  The code is yours, both COBOL
and compiler.  So is the choice.  Spend as much as you want.  
[end pitch]

We'll be working on portability issues this year, and on expanding the
supported architectures beyond x86_64 and aarch64.  (gcc-cobol requires
128-bit integer support.)   I look forward to making FreeBSD part of
that story.  

--jkl