From: Jochen Schmidt
Subject: Building CMUCL on Linux-glibc2.1 systems possible???
Date:
Message-ID: <38B3798F.45CF0B9C@gmx.de>
Hello,
Has anyone built a actual release of cmucl on a glibc2.1
Linuxsystem?
It seems to me that it is possible to run the glibc2.0
binaries, but impossible to build cmucl from source on this
systems!
Peter mentioned "binary incompatibilities" to glibc2.1.
Is it true that I have to compile cmucl with the "-export-dynamic"
linker-option if I want to do callbacks from C to Lisp?
Is this what I have to do to get the FFI variant of the GTK-bindings
to work on cmucl?
Is there a binary-release available which works (from scratch) with
this GTK-bindings?
--
mfG,
Jochen Schmidt
··············@gmx.de
Jochen Schmidt <···········@gmx.de> writes:
> Hello,
> Has anyone built a actual release of cmucl on a glibc2.1
> Linuxsystem?
The Debian packages in unstable/frozen are built on glibc2.1 (since
the upcoming Debian release will be based on glibc2.1). You should be
able to convert them to rpm's via alien, or install them by hand
(*.deb archives are really ar(1) archives with two .tar.gz archives,
one for the actual files (unpack as root from /) the other for admin
info).
Once installed, at least on Debian I've had no problems whatsoever in
rebuilding from sources (get the source tarball, too), and I've done
so frequently.
> It seems to me that it is possible to run the glibc2.0
> binaries, but impossible to build cmucl from source on this
> systems!
You can rebuild cmucl from source on glibc2.1, at least if your
running CMUCL is of the same version.
> Is it true that I have to compile cmucl with the "-export-dynamic"
> linker-option if I want to do callbacks from C to Lisp?
Yes. Recompile with "-Xlinker --export-dynamic" added to
OS_LINK_FLAGS in Config.linux and Config.linux_gencgc in src/lisp/
Recompilation is easy to do with a simple make in the main directory.
You might want to take a look at the Makefile beforehand.
> Is this what I have to do to get the FFI variant of the GTK-bindings
> to work on cmucl?
I don't know how the FFI GTK-bindings work, but if they do direct
callbacks from C to Lisp then this will usually be necessary.
Additionaly I have heard rumours that some of the GTK bindings also
require additional changes to the CMUCL sources, though I have no
first-hand experience with this.
> Is there a binary-release available which works (from scratch) with
> this GTK-bindings?
If really needed I could give you access to binaries of my patched
version of Peter's cmucl 2.4.18 Debian packages (includes bug fix for
log bug, and some local fixes I've done). But on the whole I'd
recommend doing your own rebuild, which isn't difficult to do, really
for same versions.
Regs, Pierre.
--
Pierre Mai <····@acm.org> PGP and GPG keys at your nearest Keyserver
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
From: William Deakin
Subject: Re: Building CMUCL on Linux-glibc2.1 systems possible???
Date:
Message-ID: <38B3E7BF.48CB1EAB@pindar.com>
Pierre wrote:
> You can rebuild cmucl from source on glibc2.1, at least if your running CMUCL
> is of the same version.
Sorry. I should have said this. Compiling a lisp executable for a given core
using the same source version (2.4.18 say) *is* straightforward.
Thank you for pointing this out,
:) will
From: Jochen Schmidt
Subject: Re: Building CMUCL on Linux-glibc2.1 systems possible???
Date:
Message-ID: <38B4511C.28A4E518@gmx.de>
Ok it was my error.
I've contacted Peter and his fast and helpful response
will surely help me building my first own cmucl ;)
Thanks for your help so far!
--
sincerely Yours,
Jochen Schmidt
··············@gmx.de
From: William Deakin
Subject: Re: Building CMUCL on Linux-glibc2.1 systems possible???
Date:
Message-ID: <38B3AE80.398356D0@pindar.com>
Jochen wrote:
> Has anyone built a actual release of cmucl on a glibc2.1 Linuxsystem?
Yes. Using the Debian cmucl-2.4.17 `unstable package' binary lisp
executable and safe-core I built lisp executables and small, normal and
safe cores from source for cmucl-2.4.17 through to 2.4.19 on a K6III-350
under Linux 2.2.14 using gcc 2.95.2 running with glibc-2.1.2 libraries.
> It seems to me that it is possible to run the glibc2.0 binaries, but
> impossible to build cmucl from source on this systems!
As I indicated above, it *is* possible, but not a trivial procedure.
However, if I a `numpty' like me can do it then it can't be too
difficult...
> Is it true that I have to compile cmucl with the "-export-dynamic"
> linker-option if I want to do callbacks from C to Lisp? Is this what I
> have to do to get the FFI variant of the GTK-bindings to work on
> cmucl? Is there a binary-release available which works (from scratch)
> with this GTK-bindings?
Whereof one does not know, thereof one should not speak. That is, I
don't know ;) However, if you would like more information, I would be
happy to provide them.
Best Regards,
:) will
From: Johan Kullstam
Subject: Re: Building CMUCL on Linux-glibc2.1 systems possible???
Date:
Message-ID: <m2og971k7z.fsf@sophia.axel.nom>
Jason Kantz <·····@kantz.com> writes:
> Hi,
>
> I will being making some data structures that will be manipulated as
> bits and bytes and written to a stream. I've noticed two ways to do
> this ... 1. use integers and functions like byte, ldb, and dpb
> ... 2. make bit-vectors and use sequence operations. Are there
> obvious advantages/disadvantages to either that someone can point
> out?
fwiw you can use the shift function ASH on integers but not on
bit-vectors.
--
J o h a n K u l l s t a m
[········@ne.mediaone.net]
Don't Fear the Penguin!