[apt4ssx8] Re: APT Grammar
- From: "Kent A. Reed" <kentallanreed@xxxxxxxxx>
- To: apt4ssx8@xxxxxxxxxxxxx
- Date: Thu, 21 Jun 2012 22:45:01 -0400
On 6/21/2012 2:27 PM, Brent Muller wrote:
On 6/21/2012 9:56 AM, Kent A. Reed wrote:
Gentle persons:
<...>
Toward this end, I've started a page on the APT Grammar and have
included a file containing all the syntactic productions in the Brown
et al. paper. It represents a major subset of APT 3 and will be my
basis for building an APT360 grammar.
Regards,
Kent
Some of you might have the ANSI standard that is available for the Apt
grammar. As you may know if you have it, it follows the Backus-Naur
format. When I started Aptos, I originally was going to create a
language based on this document. That got dropped when the apt360
source showed up.
Brent:
This brings up a tricky point that I have dithered around.
Because we have a working APT system in the form of APT360, I am most
interested in working on tools that know its flavor of APT. I assume
that this (APT4) flavor is best captured in the 1977 version of ANSI
X3.37 but I haven't seen that version (yet) so I can't confirm this is a
reasonable assumption. In the meantime, I think I can continue inching
forward by working from the older, partial grammar I just posted and the
contents of the APT360 manual y'all so niftily put together.
On the other hand, I can tell there are those who want to move beyond
technology that was current when Jimmie Carter occupied the White House!
I do own a copy of ANSI INCTS X3.37-1999. Frankly, I'm put off by its
"modernity" as well as its reliance on two ISO standards that are
outrageously expensive for an over-the-hill hobbyist. If I had access to
a working APT system that obeyed this version's rules I would be more
motivated to pursue it.
I mention this because I have the outlines of a parser in the Aptos
code. It already does quite a bit of the ANSI syntax, and creates a
parse tree using Yacc.
I mention on my APT Grammar page a "software tool to be named later."
It probably would not be drawn from the Lex/Flex Yacc/Bison stable just
because I remember the pain using them in the early days of STEP. I've
been looking over the current choices. Also, I'd rather use tools that
fit easily into a Python working environment because I was never a C++
coder and I've forgotten much of what I used to know about C.
For all that, Yacc/Bison may turn out to still be Hobson's choice.
Of course, this is just me bellyaching:-) Any language is the right
language if it gets the job done.
I mention this because it might serve as a starting point if we were
interested in completing or extending the grammar. I don't know that
I would have the energy to do it myself, but I would love to see some
of the modern language techniques mentioned in the Ansi standard
implemented. It might give us a starting point. It is in C++, with
the Yacc in C, of course. Whatever language we use, the Yacc file
might give us a base.
Like I said, I can tell there are others who want the latest goodies in
APT!
I guess my real concern is that Yacc heavily weights the scale toward
LALR(1) derivation and users tend to skew their grammars to suit, fixing
all those shift-reduce and reduce-reduce conflicts, for example. I'd
like to postpone the choice of derivation as long as possible. I'm
hoping to develop a grammar that can be poured into various compiler
compilers without too much handwork. This may be a fool's errand, but
we'll see.
Now that I've waved my prejudice around, I'll say that comparing what
you've already done with what I'm trying to do (and with my
understanding of ANSI X3.37-1999) would be interesting. Is your grammar
anywhere I can see it?
After learning about how apt360 works, I have an idea on how the
front-end of a new Apt compiler might be implemented. As you might
guess, the section 1 code is mostly parsing and geometry calculation;
the hard stuff is in section 2, and that is when I asked an industry
insider for some info, and he sent me the apt360 code. The separation
between the two sections is very clean, and a front-end could be
tacked on quite easily. If anyone is interested, I could go into more
detail on this.
Of course. Lay on MacDuff! I'm sure it's obvious to you that my current
interests lie more in the realm of what the APT folk called section
1---things like an APT part-program syntax verifier and geometry checker.
If I didn't say it before, I'll say it now, like you I think section 1
is just the tip of the iceberg. The other 7/8ths of the job lie out of
my sight.
As an aside, I am glad to not be the lead on this project- I am quite
happy to be a coder, and think other people could do the leadership
part better, so please don't think I would have any problems with us
starting with any part of the Aptos base for a parser.
Well, I"m not trying to be a leader here either (reminds me of Alphonse
and Gaston --- http://en.wikipedia.org/wiki/Alphonse_and_Gaston) but I
*always* have ideas, sometimes they even survive first contact with the
sun.
Do you have a place to post your aptos code base? sourceforge, the aptos
wiki, google site, etc.?
Brent.
Thanks, Brent.
Regards,
Kent
Other related posts:
