Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.postscript > #1404

Re: ps scanner - finishing touches

Followups directed to: comp.lang.postscript

Show all headers | View raw

luser- -droog wrote:

> The scanner is working and mostly complete. Re-wrote most
> of the postscript to remove local dictionaries. This should make the
> scanner re-entrant. http://code.google.com/p/xpost/source/browse/optok.c
> 
> Remaining issues: //immediate names. A naïve approach would be to add a
> variable to toke_dict to track whether we're inside a procedure (so
> immediate names are allowed) or outside (where immediate names are
> disallowed). 
Where does the PLRM state that immediate evaluated names are disaloowed 
outside of procedures????
> But grok_dict might be shared by several contexts. Another
> option is to push a new dictionary with this name. Perhaps toke_dict
> defines /in_executable_array to false, but when scanning a proc, do
> `<</in_executable_true>>begin` ... `end` around the recursive calls. This
> approach seems viable.
> 
> Another piece is ASCII85. I don't have the energy right now. :)
> 
> Another piece (that I'm putting off still, but the day looms closer) is
> Binary encodings.
> 
> But the piece standong in the way of running the interpreter
> interactively, is dealing with the file/string polymorphism in the token
> operator. I've kept this scanner code under the name `toke` as it's
> clearly not a full implementation of `token`, since it only accepts
> strings. But files can easily be read-off into strings. It's called
> buffering and happens all the time. So my thought is, to use the function
> that implements the %statementedit special file to read-off a statement
> into a string, then call `toke` on that string, and stash the remainder
> (the returned substring) into the file object's payload structure (the
> shared part), modifying all file-reading operators to read first from this
> lookahead_buffer, if present.
> 
> This appears to let me re-use a lot of existing code to get this working.
> But something about it leaves me uneasy. Adding a buffer to the file may
> have unforeseen consequences. Or so I fear.

Helge

Back to comp.lang.postscript | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

ps scanner - finite state machines luser- -droog <mijoryx@yahoo.com> - 2013-03-24 23:08 -0700
  Re: ps scanner - finite state machines, more luser- -droog <mijoryx@yahoo.com> - 2013-03-30 09:40 -0700
    Re: ps scanner - finite state machines, more luser- -droog <mijoryx@yahoo.com> - 2013-03-31 00:59 -0700
      Re: ps scanner - finite state machines, + hexstrings and names luser- -droog <mijoryx@yahoo.com> - 2013-04-22 23:33 -0700
        Re: ps scanner - finite state machines, + hexstrings and names luser- -droog <mijoryx@yahoo.com> - 2013-06-02 03:21 -0700
          Re: ps scanner - finite state machines, + hexstrings and names luser- -droog <mijoryx@yahoo.com> - 2013-06-09 12:16 -0700
            Re: ps scanner - finite state machines, + hexstrings and names luser- -droog <mijoryx@yahoo.com> - 2013-06-11 23:40 -0700
              Re: ps scanner - finishing touches luser- -droog <mijoryx@yahoo.com> - 2013-06-21 00:56 -0700
                Re: ps scanner - finishing touches Helge Blischke <h.blischke@acm.org> - 2013-06-21 11:22 +0200
                Re: ps scanner - finishing touches Helge Blischke <h.blischke@acm.org> - 2013-06-21 11:22 +0200
                Re: ps scanner - finishing touches luser- -droog <mijoryx@yahoo.com> - 2013-06-21 09:20 -0700
                Re: ps scanner - finishing touches luser- -droog <mijoryx@yahoo.com> - 2013-07-27 00:56 -0700
                Re: ps scanner - finishing touches luser- -droog <mijoryx@yahoo.com> - 2014-01-25 23:32 -0800
                Re: ps scanner - finishing touches luser droog <luser.droog@gmail.com> - 2016-11-28 08:12 -0800

csiph-web