Received: from ATHENA-AS-WELL.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA18166; Sun, 31 Jan 93 16:36:09 EST
Received: from CECI.MIT.EDU by Athena.MIT.EDU with SMTP
	id AA17222; Sun, 31 Jan 93 16:36:08 EST
Message-Id: <9301312136.AA17222@Athena.MIT.EDU>
Received: from CECI.MIT.EDU by ceci.mit.edu id AA20613g; Sun, 31 Jan 93 16:36:14 EST
To: jud@ceci.mit.edu, aybee@ceci.mit.edu
Date: Sun, 31 Jan 93 16:36:14 -0500
From: Adam Feder <aybee@ceci.mit.edu>


status report:

i changed the grammar to allow optional array dimensions and initializations
after ANY variable type.  i think that i will be able to give better warnings
(and errors) if i handle them in the actions, after accepting the erroneous
dimensions and initializations.

i fixed the lexer to recognize reals...my problem was that i wasnt quoting the
"+" in the exponent specification....+ is an operator in lex...so i got errors!

i was about to allow the '-' in the array specifications for multiple assignment
s, but i want to think about it some more...i may need to make a new type to
handle it...ie the "multipleAssignmentArrayType" and that seems odd.  i need to
ensure that the only place that gets used is on the left hand side of an 
assigment statement and i need to overload the := operator too.
is this going to be such a common occurence that we should be nice and provide
the overloaded operator?  would the fancyArrayAssign() function be as useful?
it could take an array to assign to, a list to assign from, the dimension to
increment, and the initial value for that dimension.
i also got worried because, as far as i can tell, your multiple assignment
example would cause runtime errors.  you are assigning to an array of lists,
but some of the elements in the list to assign from are strings...not lists!
(i should have caught that earlier, sorry)....in any case, i would like to have
you straighten me out if that isnt an error. while im working on the typing 
scheme i will try to figure how this fits in.  meanwhile (back at the ranch)
i will postpone putting multiple array assignments into the grammar.

(what would happen in this case?
   array(1,2,3-4-) := { 'one, 'two, 'three};
   i think -----^ will cause the parser to go into fits again....?
)

i started to put this stuff into frame and realized that i wasnt sure what
it should look like.  also, i would appreciate suggestion on naming productions
i you have any good ideas and I dont want to spend a lifetime trying to fix
both my yacc code and the frame file to reflect your suggestions.  i am
postponing importing the yacc until we have talked, but i would like to get
it taken care of soon, so that we can include it in the next DS draft to get
responses.


this message is getting long....so i'll stop.    :)

ab
