In message <3380C2B9.4D23@york.ac.uk>
          Chris Gilbert <cg110@york.ac.uk> wrote:

> Amit Gupta wrote:
> > 
> > Any chance of a RiscBSD related Wakefield show report, for those of us who
> > sadly couldn't make it?
> 
> Ok, here's a brief list of what I saw :-
> 
> Shared libs, working, will be released with gcc 2.8 which is due out in
> a month or so.

I guess you (Chris Gilbert) got this information about the expected
release date of gcc 2.8 from one of the RiscBSD chaps. I'm pretty
certain they should not have mentioned this because it is against the
FSF policy, which is not to announce releases before they are ready and
it is unpredicatable when they are ready. Thus, the FSF announces
releases only when the product is released (since at this point there is
nothing to predict).

This is important to understand because 2.8 has been due out in a month
or so for over a year (though perhaps not always `out in a month' during
the last year). If everyone had known this, just howmany `when is it
going to be released' messages do you think would be floating around the
place ?

> [optimisations for a commercial customer]
>
> I think a mention of pmap was made, and is being paid for by a company,
> so again, it's if they let it be released.

This is the sort of thing that causes me to worry about my continued use
of NetBSD. I've no objection to Mark et al doing what they are doing and
running a business with NetBSD as one of their core products, but it
upsets me that we have to live with the possibility of never seeing
these improvements. For example, the pmap code as it is now makes the
performance of the SA version very poor compared with its rated
performance.

> Big thing to note is that any code submitted to mark for putting in the
> source needs to clearly state the copyright holder(s)  This is for legal
> reasons and so that it's clear who did what :)

I would be happier if improvements to NetBSD were done on something like
a contract base for Causality who then sold it onto their customer. I'm
probably treading on dodgy ground here, especially since I don't know
the real situation, but I've always thought of NetBSD as being a free
*BSD Unix (including source) available for everyone. If a customer wants
a specially engineered release, then this is the type of service I would
be expect something like Causality to be providing, not private code
which fixes important problems in the kernel. Perhaps the customer needs
properly educating in the philosphy of NetBSD (or perhaps I do?). How do
the other NetBSD architectures work in this respect ?

In my view, this type of arrangement may well make others less likely to
put their free time into supporting NetBSD/arm32. Surely, if someone
contributes code to NetBSD, they are likely to use the standard NetBSD
copyright notice. If my understanding is correct, this gives anyone the
right to use that code, for a commercial sale if they want, providing
they keep the copyright notice on the source. No, before anyone suggests
otherwise, this is not a case of sour grapes or anything. I've always
been happy putting my spare time into NetBSD when I understood kernel
improvements would see the light of day for everyone as and when they
became available. The current situation is making me consider moving to
Linux/ARM. I really would have liked to attend Wakefield to discuss this
matter with Mark and Neil et al, but the F.A. Cup was more important to
me and the distance and travel arrangements meant it was too far.

Obviously, I would also like to say that I am grateful for the effort
put in by everyone associated with NetBSD and I would like it to
continue that way. However, I really would like clarification on the
issues I have raised above.

> What this means for RiscBSD:
> Shared libs will appear with the gcc 2.8
> Source code for the Digital NC's will be appearing some time soon.
> Digital want the source code to be available mainly so that developers
> have access to one central place for source code for the NC :)

This I like.

> Other things will come along once the legal bits of paper are signed.

This concerns me slightly, as outlined above. Perhaps Mark or Neil can
allay my fears/concerns ?

regards,
Pete

[CC: Neil and Mark @causality]