Re: push parser/lexer interface

[John W. Eaton]

In my message here:


https://mailman.cae.wisc.edu/pipermail/octave-maintainers/2013-March/032525.html

I wrote:

> I've made a number of changes to the parser and lexer recently in
> order to allow a push parser interface to really work.

That was almost but not quite correct, as character strings still
required potentially unlimited lookahead.  Now, with the following
changesets

  http://hg.savannah.gnu.org/hgweb/octave/rev/531473481084
  http://hg.savannah.gnu.org/hgweb/octave/rev/f21194531877
  http://hg.savannah.gnu.org/hgweb/octave/rev/f29dd5a7591d

that problem should be fixed and we actually can gather all input one
line at a time (even for character strings containing continuation
markers or other syntax elements that span multiple lines) and send it
to the parser in single-line chunks.

This way of gathering input and parsing opens the possibility of
having a GUI that is completely in control of all input and simply
passes the input lines it gathers to the Octave parser.

The changesets above also modify the way that continuations are
handled inside character strings.  Line continuation markers are no
longer allowed inside single-quoted character strings.  That change
improves Matlab compatibility.  Inside double-quoted character
strings, the only line continuation that is recognized is a backslash
immediately followed by a newline character.  No trailing comments or
whitespace are allowed.  These changes are noted in the NEWS file.

jwe