>>>>> "ecm" == Eric Marsden <[EMAIL PROTECTED]> writes:
>>>>> "jr" == Jan Rychter <[EMAIL PROTECTED]> writes:
 jr> In all cases, all CMUCL lisps compiled by me have a non-functioning
 jr> "directory".
 ecm> make sure you're building with the :glibc2 *feature*, to compile
 ecm> unix-glibc2.lisp instead of unix.lisp.

 ecm> More likely you are running into LFS (64-bit offset_t) problems;
 ecm> see the comment at the beginning of unix-glibc2.lisp. Possibly
 ecm> your linker isn't interpreting the hints to use the stat64 entry
 ecm> point instead of the stat entry point.

... and this was the correct hint. Thank you! Hooray hooray, after
correcting cross.lisp, I have a working lisp again!

The error I was making was to place :glibc2.1 in the list of features to
remove in cross.lisp, but to forget to remove it from the list of
features to _add_.

And this wasn't easy to find, as the resulting lisp advertised its
features as:

(:CMU19 :CMU19A :PYTHON :CONSERVATIVE-FLOAT-TYPE :MP :X86 :LINKAGE-TABLE
 :RELATIVE-PACKAGE-NAMES :LINUX :GLIBC2 :UNIX :RANDOM-MT19937 :GENCGC :PENTIUM
 :I486 :HASH-NEW :HEAP-OVERFLOW-CHECK :STACK-CHECKING :COMMON :COMMON-LISP
 :ANSI-CL :IEEE-FLOATING-POINT :CMU)

As you can see, :GLIBC2.1 isn't there -- possibly because of the line in
setenv.lisp: 

  (setf *features* (remove :glibc2.1 *features*))

thanks for all the help,
--J.


Reply via email to