1. 08 Apr, 2005 1 commit
  2. 07 Apr, 2005 1 commit
  3. 06 Apr, 2005 5 commits
  4. 05 Apr, 2005 1 commit
  5. 04 Apr, 2005 6 commits
  6. 02 Apr, 2005 1 commit
    • Christophe Rhodes's avatar
      Prolog syntax improvements: · e4ceed7c
      Christophe Rhodes authored
      * PRINT-OBJECT for PROLOG-LEXEMEs, for less pain while debugging;
      
      * OPEN-CT production rule from OPEN-CT-LEXEME, because we can;
      
      * rework the lexer a bit.  Now UPDATE-SYNTAX just invalidates the lex
        as it invalidates the parse, and UPDATE-SYNTAX-FOR-DISPLAY relexes as
        far as it needs to;
      
      * we need operator-compound-lterm and subclasses, because we cannot
        create multiple nonterminals from one rule: returning
          (make-instance 'lterm :term (make-instance 'foo ...))
        from a production rule leaves some slots in the FOO unfilled;
      
      * note my own bafflement as to why an apparently infinitely-recursive
        production doesn't recurse infinitely.  It can be fixed when needed,
        but why isn't it triggering?
      
      This version still gets various aspects of multiline lexemes wrong, but
      it's a lot better than before.
      e4ceed7c
  7. 31 Mar, 2005 1 commit
  8. 27 Mar, 2005 4 commits
    • Christophe Rhodes's avatar
    • Christophe Rhodes's avatar
      Parse [A,B]-style lists correctly. · 787cf6b4
      Christophe Rhodes authored
      787cf6b4
    • Christophe Rhodes's avatar
      "Concision is equivalent to powerfulness" · ff273ea5
      Christophe Rhodes authored
      Delete about 120 lines by defining define-prolog-rule to wrap around
      ADD-RULE.
      
      More known lacunae
        * [A,B] does not parse properly.
        * quoted tokens confuse the incremental lexer.
      ff273ea5
    • Christophe Rhodes's avatar
      Add a half-baked half-finished ISO Prolog syntax. · 2aa1652a
      Christophe Rhodes authored
      Known internal lacunae:
        * does not handle quotations correctly:
          ** fails to recognize '' as an escaped quote character (and similarly
             for other quote characters);
          ** fails to canonize quoted atoms; will fail to identify the operator
             ; from the quoted atom ';';
          ** does not contain any logic for backslash-escaped data.
        * does not implement parsing floating point numbers.
        * comments before directives get italicized.
        * way too long.  Expect some refactoring.
      
      Known suboptimal interactions with the syntax framework:
        * parses the entire buffer at every keystroke; incremental parsing caused
          a problem at some point and has not (yet) been restored.
        * displays the entire buffer at every keystroke.
        * the incremental lexer gets confused in the presence of comments.
          (see forthcoming mail to climacs-devel)
      2aa1652a
  9. 23 Mar, 2005 1 commit
  10. 20 Mar, 2005 2 commits
  11. 19 Mar, 2005 1 commit
    • Aleksandar Bakic's avatar
      Persistent/persistent-undo.lisp: new file containing the simple · 48ca0ef4
      Aleksandar Bakic authored
      version of the Undo protocol that relies on persistent buffers. Marks
      are not persistent (yet?), they are only "fixed" to prevent trivial
      errors.
      
      climacs.asd: added Persistent/persistent-undo.lisp.
      
      gui.lisp: added calls to full-redisplay at the end of com-undo and
      com-redo (could not find a better way).
      
      pane.lisp: added copyright info; fixed a bug in with-undo macro;
      modified extended-binseq2-buffer to inherit from p-undo-mixin; cleaned
      up some mark cloning code.
      48ca0ef4
  12. 18 Mar, 2005 1 commit
    • Robert Strandh's avatar
      Added setf methods for offset of parse-trees. Either a numerical · 236611db
      Robert Strandh authored
      offset can be given, in which case, the start-mark must exist (since
      we don't know the buffer), or else a mark can be given, in which case
      it is cloned.
      
      Removed references to start-mark from html-syntax.lisp, and removed it
      from the export list of the climacs-syntax package.
      236611db
  13. 17 Mar, 2005 1 commit
  14. 16 Mar, 2005 2 commits
  15. 15 Mar, 2005 5 commits
  16. 13 Mar, 2005 2 commits
    • Aleksandar Bakic's avatar
      Line-oriented persistent buffer (binseq2). Warning: Need to fix minor · 94f6939a
      Aleksandar Bakic authored
      bugs (related to number-of-lines-in-region, I believe).
      
      base.lisp: Added faster methods on previous-line, next-line,
      buffer-number-of-lines-in-region.
      
      pane.lisp, cl-syntax.lisp, html-syntax.lisp, text-syntax.lisp:
      Replaced some calls to make-instance to calls to clone-mark and (setf
      offset), in order to avoid passing climacs-buffer to marks. This also
      made possible to get rid of delegating methods on syntax.
      
      climacs.asd: Added Persistent/binseq2.
      
      packages.lisp: Added binseq2-related symbols.
      
      Persistent/binseq.lisp, Persistent/obinseq.lisp: Cleanup.
      
      Persistent/persistent-buffer.lisp: Added code for binseq2-buffer and
      related marks. Also some minor fixes.
      94f6939a
    • Robert Strandh's avatar
  17. 11 Mar, 2005 3 commits
  18. 10 Mar, 2005 1 commit
    • Robert Strandh's avatar
      More progress on html-syntax, which may eventually become a model · 5c330694
      Robert Strandh authored
      for many different language syntax modules.
      
      The display function now traverses the parse tree up as long as a
      valid parse tree exists.  The rest of the display is done from the
      token sequence.  It is likely that all of this can be abstracted out
      and put into syntax.lisp so that html-syntax would just become a
      client among others for this traversal.
      
      Not only is the cursor still not displayed, whitespace is not handled
      during the traversal of the parse tree.  This will likely be fixed in
      the next few day.
      5c330694
  19. 07 Mar, 2005 1 commit
    • Robert Strandh's avatar
      First attempt at a display function (for html syntax) that uses the · 4f8663ce
      Robert Strandh authored
      output of an incremental lexer and parser.  This code is not complete
      yet:
      
        * right now, it uses only the lexer output, and not the parser
      
        * the cursor is not displayed yet
      
        * it is too slow, most likely because the output records are all in
          one big bag, as opposed to being tree structured, such as the
          lines of basic syntax.
      
      The slowness is not a serious problem, because one day, McCLIM will
      have tree-structured output records itself, and because most syntax
      modules (including this one, very soon) will have some tree structure
      itself.  It might be worthwhile to display the part of the buffer
      beyond a parse error in some artificially structured way, such as by
      lines as in the current basic syntax.
      4f8663ce