[Coco] Entire OT: Me and mine

John R. Hogerhuis jhoger at pobox.com
Thu Jun 2 12:52:31 EDT 2005

On Thu, 2005-06-02 at 06:15 -0500, Boisy G. Pitre wrote:

> I'm sounding harsh here, but frankly, I don't see how you can be a  
> part of several high profile projects and just skip and run out for  
> "the next couple of years" without being called on it.  Then, to be  
> frankly honest, I don't see much work for the talk that you've been  
> doing.
> Boisy

Well what work James did on sdcc he has already checked in. Basically he
modified the register allocator for 6809 and started on the code
generator that uses it. 

The way I see it, SDCC is a free software project -- people contribute
when they can, where they can, and everybody is encouraged to jump in at
any point. People start things all the time; it doesn't cost anyone
anything to say "I'm working on this." But since no one is getting paid,
other duties and issues distract. This is no professional commitment he
made, so even if one discounts the "burnout" (I won't because it is real
and I've experienced it myself), he still doesn't owe anyone anything.
Free software development is a party -- you show up because it's fun and
because you want to come. Development right now isn't fun for James, and
I don't see any point in harassing him for not coming to the party.

For the most part with free software it's all about "show me the code."
James knows that. Until something tangible/testable exists, it's just
vaporware in both the free software world, and the proprietary software

I knew he seemed to be stuck for whatever reason sometime back, so I was
trying to get everyone to pitch in to discussing it to keep things
moving. That didn't really work.

As I've said on list I've been sizing up the code generator trying to
get a handle on it; James was taking the brute force approach of trying
to do the mapping of hc08 to 6809. Looking at the code generator (w/
ralloc.c it's on the order of 8000 lines of code) I think it needs
serious rework: it's not modular enough for my taste, it's repetitive...
just plain not well-factored enough to build on. I think it's useful as
a model, but we really need a 6809 code generator and the HC08 is
different enough and the code needs to be reworked enough that I think
we should probably use the hc08 port and James' work on it as a
reference rather than directly modifying it.

Of course anyone who is interested is welcome to take whatever approach
seems best to them. What works first seems to be the way things usually
go :-)

So I say good luck to James, and that cc6809 project needs help on the
sdcc code generator, and the standard library.

-- John.

More information about the Coco mailing list