Re: RFS: open-axiom
On Sat, 27 Aug 2011 23:01:12 +0400, Игорь Пашев <email@example.com> wrote:
> I have fixed all lintian issues, except two:
> 1. image-file-in-usr-lib (e. g.
Hmm. This should be fixed eventually, but if it will be fixed upstream
anyway, it is probably more sensible to wait for that fix. In the
meantime, I think we can live with it. I know that racket had a few
hundred of these before I did the symlink trick. Maybe others feel
differently about the severity of this problem.
You should probably investigate whether the fasl files are architecture
independant (i.e. can work on arm, mips, sparc and friends). In that
case, it will be worthwhile splitting out an open-axiom-common
architecture independent package. This only needs to be build (and
stored) once. On the other hand, the fasl files may depend tightly on
the machine architecture.
> 2. unusual-interpreter + script-not-executable (e. g.
> The second is a feature of SBCL: *.fasl files are compiled lisp code,
> and they must have a special header. Changing shebang makes SBCL core
> contains an illegal byte in the FASL header at position 0: Expected 35, got
The literal error message complains because the file does not start with
#. Maybe you have a blank line before # FASL? I'm no common lisp expert though.
As far as I understand, the #! is added by sbcl to permit fasl files to
be run as scripts. But this is not much use if the path inserted is in
the build tree, namely something like:
I'm not sure if open-axiom uses this feature (and I notice sbcl itself
also has broken #! lines in it's fasl files. In your case, I suspect the
upstream build process is putting the wrong path into the #!
lines. interpsys isn't installed by debian sbcl and it isn't included in
the open-axiom source package.