>
> I think you really know what you're talking about. I agree with your
> remarks above, especially points 1 and 2 above, and your remarks about
> the difficulty of building GCC itself and having different coexisting
> GCC's at once are exactly right.
>
> Shall we start taking some genuine steps toward doing what you propose
> (which is very close to what I've proposed recently in a blog post)?
>  I don't have the time to work on this much right now, but would
> definitely like to strong encourage such work.
>
>  -- William

Time is my problem, too.

I want to compare (at least) three settings on my system: 1) a
"native" build of the Sage distribution, 2) Sage-on-Prefix (i.e. a
minimal Gentoo Prefix install, plus the Sage-on-Gentoo overlay
installed in its prefix incarnation), and thirdly 3) the Sage
distribution built on top of a minimal Gentoo prefix install, i.e.
using that one's ("hosted") gcc, not the system gcc, etc.pp.
Till now, 1) works (as it did for the past years), and for 2) I got
down to just only 17 packages still needed to be tweaked/installed to
have Sage-on-Prefix working, the stoppers currently being polybori
(doesn't compile for some reason, some C++ template stuff blowing up)
and ntl (nothing essentially Prefix or Sage specific --- matter-of-
factly the base system I use, OS X 10.4 Tiger, has its usual "install-
name" and ".dylib versus .so" problems, and the ntl source code does
not build out of the box there, so I have to move some more patches
from the spkg to the ebuild).
As for 3), I didn't start it yet, I expect problems to detect properly
the underlying system (which shall not be the "host OS" in this case,
but the "hosted" Gentoo Prefix) scattered all over the place, but
nothing "really serious".

Having these three alternatives up and running, one could say more
about the feasibility and the future of these approaches.

On Windows, I'm thinking about a fourth alternative (which is not at
all Windows specific): 4) put a kind of base Linux distribution (e.g.
the Ubuntu server one specialized for this) inside some VirtualBox
image, together with guest additions, then (this is the important
part) permanently share some folder of the host system's file system
inside the virtual image, and then install there all the files (native
Sage distribution, or Sage-on-Prefix, ...).

Importantly, neither during "pure usage" nor during development work
on Sage, files *inside* the virtual image are to be be written/
altered; each startup of Sage shall use the same "static" virtual
image (and its files, configuration, ...). This would address one big
hurdle of "Sage on Windows by virtualization", namely that you have to
maintain your files, changes, backups "in and out" of the virtualized
system. Having all these files (i.e. the whole Sage file system
subtree, as well as the "home/dotsage" folder) directly on/inside/
accesible by the (Windows, or whatever host OS) file system, one can
make backups, use editors, ... just as one is used to, directly on/by
the host OS.

In the VirtualBox 3.x.y series, there was/is a bug (#818 in their
database) that prevents this, because symlinks don't work with shared
folders (I found that out the hard way, trying to install Sage into
such a shared folder ...). With VirtualBox 4.0.0. beta, this bug is
announced to be fixed, but I didn't do any testing/installing yet.
>From a distance, this is also is nothing else but "moving up or down"
the line between where your "host OS" ends, and where your "hosted
distribution" starts. The goodie would be that on native Windos 64-bit
systems, one could use 64-bit powered Sage (with only a small overhead
of less than 10 percent due to the virtualization, and the shared
folder access mechanisms), whereas e.g. Cygwin is (and will be for the
foreseeable future) 32-bit bound.

Cheers,
Georg


P.S.:
Francois, I thought you were "on leave" moving to a different
location? Nice to see that you read Sage threads nevertheless!

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Reply via email to