Skip to content
Snippets Groups Projects
  1. Jul 13, 2006
  2. Jul 12, 2006
    • rtoy's avatar
      The subtype_Names array had one extra slot, so that · 8932f44e
      rtoy authored
      closure_function_header and the following entries were not correct.
      
      Also, rename unused function header 1 and 2 to be "byte code function"
      and "byte code closure", respectively, since that's what internals.h
      says.
      
      We really need to generate this list during genesis or something
      instead of having to do this by hand!
      8932f44e
  3. Jul 10, 2006
  4. Jul 07, 2006
  5. Jul 05, 2006
    • rtoy's avatar
      Fix bug about LISP::SOCKET-ERROR not being a valid class. From Craig · 1b8f6914
      rtoy authored
      Brent Ludington, cmucl-help, 2006-06-21.
      
      bootfiles/19c/boot-2006-06-3.lisp:
      o Bootstrap file to remove LISP::SOCKET-ERROR in favor of
        EXT:SOCKET-ERROR.  (Not really needed, if you just answer the
        restarts in the obvious way.)
      
      code/exports.lisp:
      o Export EXT:SOCKET-ERROR.  The LISP package already uses the EXT
        package, so we're set.
      1b8f6914
  6. Jul 01, 2006
  7. Jun 30, 2006
  8. May 31, 2006
  9. May 30, 2006
  10. May 23, 2006
  11. May 20, 2006
  12. May 18, 2006
  13. May 15, 2006
  14. May 11, 2006
  15. May 09, 2006
  16. May 07, 2006
    • rtoy's avatar
      ppc/arith.lisp: · b07ae7fb
      rtoy authored
      o Comment out the +/fixnum and -/fixnum.  The C code doesn't support
        the fixnum-additive-overflow-trap, and this vop doesn't do enough
        checking of the args.
      
        (defun foo (x y) (the fixnum (+ (the fixnum x) (the fixnum y))))
      
        triggers bad behavior if there's an overflow, or if x or y is not a
        fixnum.
      
        Everything still works without these vops.  The check for fixnum
        args becomes explicit, and the result is explicitly checked for a
        fixnum result.
      
        I'm guessing this was modeled after the Sparc version of these vops,
        which actually checked the tag bits and for overflow.
      
      ppc/parms.lisp:
      o Remove the fixnum-additive-overflow trap because we don't use it
        anymore.
      b07ae7fb
  17. May 03, 2006
  18. May 01, 2006
    • rtoy's avatar
      Update from logs. · b00064f8
      rtoy authored
      b00064f8
    • rtoy's avatar
      code/format.lisp: · ef77f543
      rtoy authored
      o Add an additional case where we want to print out a trailing zero:
        There's no width constraint and the previous character was a decimal
        point, so the fraction to be printed is zero.
      
      code/print.lisp:
      o Honor the d option if we run out of room so (format nil "~,2f"
        0.001) produces "0.00", not "0.001".
      ef77f543
  19. Apr 28, 2006
  20. Apr 26, 2006
    • rtoy's avatar
      src/save.lisp: · 29b5082e
      rtoy authored
      o Add new search-list "ld-library-path:" which contains the value of
        the environment variable "LD_LIBRARY_PATH".
      
      src/foreign.lisp:
      o Modify LOAD-FOREIGN so that we will try to load a single file as a
        shared library first.  If that fails, we try loading it as an object
        file.  We do not try to see if the library exists (via probe-file or
        anything).  Instead we let dlopen do whatever it would normally do
        to find the file, including searching LD_LIBRARY_PATH.
      
      o REINITIALIZE-GLOBAL-TABLE now has a few restarts to allow the user
        to decide what to do if a shared library cannot be found.  The
        restarts are ignoring the problem, trying to reload the file again,
        or specifying a new path.
      
      o REINITIALIZE-GLOBAL-TABLE is placed on
        EXT:*AFTER-SAVE-INITIALIZATIONS* now.
      29b5082e
  21. Apr 13, 2006
  22. Apr 07, 2006
  23. Apr 03, 2006
  24. Mar 18, 2006
  25. Mar 17, 2006
    • rtoy's avatar
      lisp/Linux-os.c: · 7575e738
      rtoy authored
      o Add function get_h_errno to extract h_errno.
      
      code/internet.lisp:
      o On Linux, call the get_h_errno function to get h_errno instead of
        calling __h_errno_location ourselves since that's not part of the
        public API.
      7575e738
  26. Mar 14, 2006
    • rtoy's avatar
      Fix a few compiler warnings pointing to real issues: · c4cd431c
      rtoy authored
      o The error message in CONNECT-TO-INET-SOCKET could be getting the
        wrong message if closing the socket causes an error.
      
      o The function (FLET DOTTED-QUAD CONNECT-TO-INET-SOCKET/NON-BLOCKING)
        was not using its argument.  (No actual change in behavior because
        all uses of DOTTED-QUAD were with an arg of ADDR anyway.)
      c4cd431c
    • rtoy's avatar
      o Add docstring for *IGNORE-WILDCARDS*. · 172f99d7
      rtoy authored
      o Make UNPARSE-UNIX-PIECE honor *IGNORE-WILDCARDS*.
      
      Requested by Lynn Quam so namestrings can be given to run-program and
      friends without pathname wildcards escaped.
      172f99d7
    • rtoy's avatar
      Linux (just x86?) doesn't have h_errno. But it has __h_errno_location · 6b5b3c08
      rtoy authored
      to return the address of the (thread-local) h_errno variable, so use
      that to get the value of h_errno.  Add GET-H-ERRNO function to return
      the value of h_errno.
      6b5b3c08
  27. Feb 27, 2006
  28. Feb 25, 2006
Loading