Re: gnumach release

[Roland McGrath]

> I think that we should try to fix the support for a large amount of memory. 
> Somebody posted a work around today to debian-hurd, and as it is in the
> Linux code, maybe it is not too difficult to find the real fix by looking at
> some newer version of this Linux code.

Ok.  That workaround doesn't make sense to me, and I don't know any details
about exactly what the problem is when it comes up.  The code in question
is in Shantanu's or Okuji's support code for the Linux drivers, not in
copied Linux code per se.  So there is no obvious parallel to compare to.

> I have made a huge progress with mach.texi, which is now 157kb large and

Excellent.  I didn't even know you had been doing this (or anyway I
forgot).  But I don't think we should worry about this for the release.
We might as well just release the documentation separately whenever it is
ready, under the FDL if possible.  

Not only is that interface documentation in fact not specific to gnumach vs
oskit-mach, but it is still correct for any pre-OSF CMU Mach 3.0 offshoot
(maybe including xmach, I don't know what their base is).

If you also had compilation instructions for gnumach in there, it would
make sense to break those out into a separate install document that goes
into the gnumach tree, and we could put that into the 1.3 release if you
have something.

Ok, I've just downloaded your file and looked at it.  Very GNUish.  Well,
that would be good to have in there.  I dunno.  Hack on it if you feel like
it.  But I'd say we shouldn't worry about that for the 1.3 release (it's
never had any documentation before).  You can put that up separately as
it's ready, and we can put an htmlized version in the www.gnu.org hurd
pages.  You should post something to the lists about that soon and get some
volunteer help in finishing it up; I think it will be very popular (well,
on the Hurd scale of popularity :-).

The obvious addition it could use is documentation of the boot script
syntax, which has never been really specified anywhere before.

I'm inclined to suggest we separate this into a few different manuals.  Or
at least separate texinfo files that might or might not be combined into
one book.  One manual purely about the programming interfaces--that one
will remain the same for several variants; this could also cover mig and
other tools if/when that gets described.  Another manual for installation
and configuration, which would cover booting issues, boot scripts, etc;
that much is common to gnumach and oskit-mach.  That might also be the
place to describe configure options for compiling and so forth, the section
from which the INSTALL file can be generated; those of course differ
substantially for every variant.  

I dunno how to organize it.  But I think if you solicit some volunteers you
will get plenty of help on this.

When we get this manual in shape, the FSF will probaby be interested in
printing it up in small runs to add to their catalog of hardcopy manuals
for sale.  (And then I could get a hardcopy to use nicer than these old
bound photocopies Thomas made for us in 1991 or so. :-)