Rex Dieter <rdieter1@...> writes:
> Christophe Rhodes wrote:
>> Rex Dieter <rdieter1@...> writes:
>>> In my efforts to get sbcl included in Fedora Extras, we'd like to
>>> support the linux-powerpc platform, but I need a
>>> newer/recent/current binary from which to bootstrap. Currently,
>>> 0.8.15 is the latest downloadable from sourceforge.
>> get a newer binary, I'd start by getting a recent .deb -- which is an
>> ar(1) archive -- and extracting sbcl and sbcl.core from that.)
> Not an option. I need official/upstream bootstraps, or we can't support
> the platform.
Who gets to decide what is official and what isn't? Why is (say) my
word better than (say) Kevin's or Peter's? In fact it can't be, I'm
not an administrator for SBCL; if you want official official blessing
for any given binary, you'll need to talk to Bill Newman.
The only conceivable thing that I can possibly think that you, or your
masters, are attempting to guard against by this wanting an "official"
"upstream" bootstrap is an attack of the _Reflections on Trusting
Trust_ form. In which case, simply cross-compile your first binary
yourself from (say) openmcl, with which there is little developer
overlap... oh but wait, we could be colluding with clozure.com to hide
the recompiling magic, so that doesn't work.
Here's a way forward: you could fork SBCL; you could call it "Steal
Bank Common Lisp", to reflect the _true_ way that Carnegie and Mellon
made their bucks, and then act as your own official upstream and
provide your own binaries.
Or, here's something which could be fun: you could build clisp from
source, fix the bugs in it which prevents the current version of that
from building SBCL, and then build SBCL from that clisp. Then
assuming that Ken Thompson and I haven't been extremely devious, you'd
be safe. You could make yourself safer by instead fixing enough bugs
in gcl to build SBCL; then you'd be certain that no-one has been
putting in nasty stuff. Oh, wait, you've only got some random
anonymous dude's e-mailed word for that; no dice.
A possibly simpler way to progress is to use whatever x86 version you
have (I presume you have an x86 version) to build a powerpc version.
But, frankly, the need for "official / upstream bootstraps" sounds
like just the kind of administrative stupidity that I would suggest
you try to get away from; whether that's by openly questioning that
policy to get it changed, or supporting a less insane distribution --
or whether you ignore my advice -- is of course up to you.