Skip to content
Snippets Groups Projects
  1. Sep 03, 2010
    • rtoy's avatar
      More fixes. With these changes, the normalization and word-break · 68f0165a
      rtoy authored
      tests pass without problems.  Copying the lines from the test files to
      a new file also produces a file that is identical to the original.
      (Previously, this was failing.)
      
      o For debugging, clear out the parts of in-buffer that have already
        been processed.  Not enabled normally.
      o When retrying the read to get more octets, we weren't reading the
        new octets into the correct spot in the in-buffer, and we weren't
        updating the in-length with the correct length.
      68f0165a
    • rtoy's avatar
      fix some bugs in previous commit. · 82c7c2ef
      rtoy authored
      o For debugging, the last sbuf element is
        lisp-stream-string-buffer-len, not one more than that.
      o We were not copying the last element of the string-buffer to the
        beginning.
      o Forgot to set the stream in-length correctly when we had to retry
        reading octets to get a whole character.
      82c7c2ef
  2. Sep 02, 2010
    • rtoy's avatar
      Fix yet another bug in the FAST-READ-CHAR-STRING-REFILL. This shows · ef87b195
      rtoy authored
      up when running the word break test in
      i18n/tests/word-break-test.lisp.
      
      extfmts.lisp:
      o Return the number of characters that were actually converted instead
        of the position of the starting point of the output string.
      
      stream.lisp:
      o In FAST-READ-CHAR-STRING-REFILL, sometimes, we'll only read one
        octet into the octet buffer, and the octet will be the first octet
        of a multi-octet character.  If this happens, we need to try to read
        some more octets in so that the call to FAST-READ-CHAR-STRING-REFILL
        can return a character.  We only retry once.  If this still fails to
        read enough octets to form a character, we're hosed since we don't
        check for this.  (Should we?)
      
        Need to refactor this code a bit too.
      ef87b195
    • rtoy's avatar
      Merge from HEAD, adding -w option. · 0dc06545
      rtoy authored
      0dc06545
  3. Aug 28, 2010
    • rtoy's avatar
      code/x86-vm.lisp: · 91e35835
      rtoy authored
      o Fix SIGCONTEXT-FLOAT-REGISTER to handle SSE2 better.  In particular,
        single-floats were incorrect for XMM registers.  Also update this to
        handle complex double-floats and complex single-floats.
      o Update %SET-SIGCONTEXT-FLOAT-REGISTER to match
        SIGCONTEXT-FLOAT-REGISTER.
      
      general-info/release-20b.txt:
      o Update
      91e35835
  4. Aug 26, 2010
  5. Aug 25, 2010
  6. Aug 24, 2010
  7. Aug 18, 2010
  8. Aug 15, 2010
  9. Aug 14, 2010
  10. Aug 13, 2010
  11. Aug 09, 2010
  12. Aug 06, 2010
  13. Aug 05, 2010
  14. Aug 04, 2010
  15. Aug 03, 2010
  16. Aug 02, 2010
  17. Aug 01, 2010
Loading