BABYL OPTIONS:
Append: 3
Mail: SRC:<SCHEME>SCHEME-BUGS.TEXT.0
Version:5

1, bad-header,,
*** EOOH ***

Date: 18 Oct 1982 1613-EDT
From: W. Scott Meeks <LS.WSCOTT at MIT-EECS>
Subject: Misleading example
To: Bug-scheme at MIT-EECS

   	There is what I consider to be a misleading example in the Scheme manual's description of eq?; the example (eq? 1000 1000) returns nil, implyingthat eq? of any integer and itself will return nil.  This is not true of any
integer less than 1000, so this example seems misleading.-------


Date: Sunday, 17 October 1982  11:31-EDT
Sender: GJS at MIT-OZ
From: GJS at MIT-MC
To:   DF.WANUGA at MIT-EECS
Cc:   bug-scheme at MIT-EECS
Subject: LET
In-reply-to: The message of 14 Oct 1982  21:32-EDT from DF.WANUGA at MIT-EECS

ugh.
You are right.  Unsyntaxer is trying hard to print it out as a legit program.
I suppose it is time to really try to fix it.!
-------


Date: 14 Oct 1982 2132-EDT
From: DF.WANUGA at MIT-EECS
Subject: LET
To: bug-scheme at MIT-EECS

(cons 'let (cons 'a '(b c)))  ==>  (let () b c)

(cons 'letn (cons 'a '(b c)))  ==>  (letn a b c)

How come?
-------


Date: 13 Oct 1982 2131-EDT
From: Patrick A. O'Donnell <PAO at MIT-EECS>
Subject: Line length
To: bug-scheme at MIT-EECS

Scheme fails whenever what it wants to print is wider than the screen.
For instance in the trace output, if the result from a function is so wide
that the function cannot be printed, then Scheme prints an unbounded number
of spaces.
		- Pat
-------


Date:  9 Oct 1982 1310-EDT
From: Patrick A. O'Donnell <PAO at MIT-EECS>
Subject: Line lengths...
To: bug-scheme at MIT-EECS

Scheme cannot handle degenerate terminal widths (e.g. 0) nor long
line lengths (e.g. 130) > 127 chars.  Maclisp has a problem with LINEL
for wide terminals.  (I'll look into that problem.)

Mode of lossage: Prints an unbounded number of spaces while attemting
to display its startup message.
		- Pat
-------


Date:  5 Oct 1982 2043-EDT
From: Mark A. Hamilton <LS.MAH at MIT-EECS>
Subject: Notes and Huffman Trees
To: bug-scheme at MIT-EECS


	I the future, the notes on Huffman Trees should reflect the fact that
not all trees are one sided, but that they in fact branch both ways.  I 
realize that the algorithm given leads to this conclusion, but an example
would make things infinitely more clear.

Thanx.

ls.mah

-------


Date:  4 Oct 1982 1536-EDT
From: Philip A. Earnhardt <LS.PAE at MIT-EECS>
Subject: ^g problems
To: bug-scheme at MIT-EECS

If one gets an error in reading a file, goes back to the editor (while
still in the error loop), fixes the problem, goes back to scheme
and types ^g, scheme will get an error trying to delete the
_SCHEDIT file (when trying to unwind to top level). A second ^g seems
to be able to return to top level.

Perhaps a reasonable thing to do would be to have the edit function 
send you to top level before it does anything. Would this screw anyone?

phil
-------


Date: Sunday, 3 October 1982  22:57-EDT
Sender: HANSON at MIT-OZ
From: HANSON at MIT-MC
To:   Richard Mark Soley <Soley at MIT-OZ>
Cc:   bug-scheme at MIT-OZ
Subject: Scheme printer bug.

I'm not claiming that the behaviour of the SCHEME printer is correct.
In fact, there is (I believe) a clearly defined line at which it is
known that the object being printed is a representation of SCHEME
code.  Unfortunately, I haven't had much time to fix it; the hour I
have spent so far has been insufficient.
-------


Date:  3 Oct 1982 1956-EDT
From: Thomas A. Russ <DF.TAR at MIT-EECS>
Subject: Printer bug?
To: bug-scheme at MIT-EECS
cc: df.tar at MIT-EECS

There seems to be a bug in the Scheme printer.  I get the following
rather unusual performance: (see arrows)

(*PRINT-DEPTH* and *PRINT-BREADTH* are bound to NIL)


==> (translate lform)

(LET (LET
      (FACT '2)
      (FACT FACT))					<==========
  (FACT
   (LAMBDA (FACT)
     (LAMBDA (N) (IF (EQ N '0) '1 (MUL N (FACT FACT (SUB N '1)))))))) 


==> (cadr (%out))
(LET (FACT
      (QUOTE 2))
  (FACT FACT FACT))					<==========

-------


Date:  3 Oct 1982 1557-EDT
From: Richard Mark Soley <Soley at MIT-OZ at MIT-MC>
Subject: Scheme printer bug.
To: bug-scheme at MIT-OZ at MIT-MC

It's all well and good that the printer knows about Scheme structure,
but not so great when you have a whole class (6.847) writing scheme
code to interpret a non-scheme Lisp-like language.  There should
at LEAST be a printing function which doesn't print garbage.

	-- Richard
-------


SOLEY@MIT-MC 09/29/82 23:23:57
To: (BUG SCHEME) at MIT-MC
Oy! I could kill!  There is a bug in the Scheme printer.
Try it yourself:

==> (define a `(let (let ((a b) (c d) (e f) (g h)) (quux)) (mumble)))
A

==> A 
(LET (LET ((A B) (C D)) (QUUX)) (MUMBLE))

Now, I spent a LONG time with this, sure that it was a backquote
bug.  However, when I tried it with the backquotes hand-expanded,
it still happened.  Yes, indeed:

==> (cadr a)
(LET ((A B) (C D) (E F) (G H)) (QUUX))

You got it: some wort of strange bug in the printer.  I don't CARE
that it's not good Scheme - is the printer checking out things I PRINT?

	-- Richard Soley


Date: Wednesday, 29 September 1982  03:19-EDT
Sender: HANSON at MIT-OZ
From: HANSON at MIT-MC
To:   bug-scheme at MIT-OZ


I have fixed the indentation of FLUID-LET and NAMED-LAMBDA in the
SCHEME emacs init file on OZ:<EMACS>.  Whoever knows how to transfer
this new file to EE (and has the appropriate access rights), please do
so.  Thanks,
			Chris
-------


Date: 21 Sep 1982 2006-EDT
From: Mark A. Hamilton <LS.MAH at MIT-EECS>
Subject: Unknown break
To: bug-scheme at MIT-EECS


   A student had his scheme break due to an error in photo.

   He typed (photo 'hhh), and it never recovered.  It would accept input,
but not evluate it, and after a ^C, a contiue left him in a state where
nothing moved, ^G did nothing, etc.

  The image has been saved as <ls.mah.scheme>scheme.sav, and the job
detached (job 23, <ls.bry>) if it's still there when you read this...

ls.mah

-------


Date: 18 Sep 1982 2344-EDT
From: Mark A. Hamilton <LS.MAH at MIT-EECS>
Subject: Re: popular bugs
To: UC.MP at MIT-EECS
cc: 6.001-staff at MIT-EECS, bug-scheme at MIT-EECS
In-Reply-To: Your message of 18-Sep-82 1602-EDT


	It would be better if ^X^Z acted like ^XZ.  Most students are 
just confused about the scope of control, and don't understand that there
is a difference.

	Mark (ls.mah)

-------


Date: 18 Sep 1982 1557-EDT
From: Mark Plotnick <UC.MP at MIT-EECS>
Subject: popular bugs
To: bug-scheme at MIT-EECS, 6.001-staff at MIT-EECS

Here are some problems that 6.001 students have been having:

- they are typing backspaces in Scheme to rubout characters.  The
backspaces are kept as parts of atom names, and of course these
atoms turn out to be undefined.  Since Scheme is reading a
character-at-a-time from the tty anyway, it shouldn't be too hard to have it
treat BS the same as a DEL.
This problem came up a few years ago in 6.031 on the PDP11.  I remember
Ed Gilbert going around to all the DecWriters and prying all the BS keys
off.

- they are typing ^X^Z to get out of an emacs that was invoked via (EDIT).
Thus, the changed file isn't transmitted to Scheme.  As long
as you're changing key bindings in the emacs that's invoked
by (EDIT), you might as well rebind ^X^Z (and whatever return-to-superior is)
to behave the same as M-O (zap buffer to scheme).

	Mark
-------


Date: 17 Sep 1982 1414-EDT
From: DF.TAR at MIT-EECS
Subject: LEDIT connection.
To: bug-scheme at MIT-EECS
cc: tar at MIT-ML

	The connection between SCHEME and EMACS via LEDIT seems to me to 
be incomplete.  If I am in scheme, and type (EDIT), then the job switches
to emacs, as it should.  However, I don't seem to be able to return to
scheme from the editor (via M-C-Z).
				-- Tom Russ
-------


Date: 16 Sep 1982 1950-EDT
From: Stephen E. Robbins <LS.STEVER at MIT-EECS>
Subject: Debugger lack of feature...
To: bug-scheme at MIT-EECS


Suggestion: Have some way for the debugger to be able to break and
trace procedures in lower environments (there may be a way to do it
but the TA here doesn't know it if there is one).  Fr'instance:

==> (pp fib)
[PROCEDURE (FIB N)
    (DEFINE (FIB-ITER N-1 N-2 N)
       (IF (= N 0) N-1 (FIB-ITER (+ N-1 N-2) N-1 (- N 1))))
    (FIB-ITER 1 0 N)]

==> 

Suppose there were a bug in FIB-ITER, any way (short of inserting a
(BREAK) within FIB after the DEFINE) of having the debugger stop
once the FIB environment is set up to let me set a TRACE on fib-iter?

			Stever

-------


Date:  8 Aug 1982 2027-EDT
From: Hazel Rovno <ZVONA.HAZEL at MIT-OZ>
Subject: scheme doesn't
To: bug-scheme at MIT-OZ
cc: zvona at MIT-OZ

@scheme

? Error in SHARE boot program: File not found
-------

Date:  7 Jul 1982 0956-EDT
From: Joshua D. Marantz <UC.JOSHM at MIT-EECS>
Subject: ;I/O ERROR DURING OUTPUT
To: BUG-SCHEME at MIT-EECS

WHEN ONE MAKES SEEMINGLY TRIVIAL TYPOS WHILE IN SCHEME, ONE GETS AN
INFINITELY REPEATING ;I/O ERROR DURING OUTPUT FOLLOWED BY SEVERAL LINES.
THE ONLY THING THAT ABORTS THIS ON EE IS ^C.  ONE EXAMPLE OF THIS IS
(TO THE ---> PROMPT) TYPING A PERIOD AND THEN A SPACE.  CAN SOMEONE FIX
THIS READER BUG?  IT DOESN'T SEEM T HAPPEN IN MACLISP
 					-JOSH
-------

Date: 1 May 1982 09:51-EDT
From: Hal Abelson <HAL at MIT-MC>
Subject: maybe a bug?
To: LS.ZZZ at MIT-EECS
cc: BUG-SCHEME at MIT-MC


Yup, its a bug alright.  (Actually a misfeature, do to the way that we
have embedded Scheme in the underlying MacLisp.)  The problem is that
although CAR and CDR work on streams, some of the other list structure
primitives do not, even though, conceptually, they are defined in
terms of CAR and CDR.

LENGTH is one of the primtives that does not work.  It is not smart
enough to recognize when it needs to UNDELAY part of a stream.

What's happening to you is that the argument printing as part of the
traces causes UNDELAYing, which changes the answer returned by LENGTH.

Here's another example:

(define x (cons-stream 1 (cons-stream 2 (cons-stream 3 (cons-stream 4
(cons-stream  5 (cons-stream 6 nil)))))))

;;a stream of 6 elements

(length x)

3

;;huh? -- length is looking at the internal representation of the
;;stream, which contains a pointer off to the delayed part, which is
;;not expanded.

(print-stream x)
{1 2 3 4 5 6}

;;now print-stream has forced the expansion.  This, by the way, has
;;CHANGED the internal representation of x (remember those
;;call-by-need thunks?)

(length x)
12

;;the answer given by length is still wrong, anyway, since it doesn't
;;take account of the factor of 2 overhead for streams.

Moral: For us, we need to fix the implmentation.  (We knew this bug
was there, but didn't think it would rear its head.)

For you, don't use the primtive LENGTH to count the number of items in
a stream.

Thanks for finding the bug.


Date:  1 May 1982 0754-EDT
From: Michael J. Konopik <LS.ZZZ at MIT-EECS>
Subject: maybe a bug?
To: bug-scheme at MIT-EECS

I have a small query of my own to make here...

I was fooling around with the query system in ps11, and to find
out some of the characteristics of the output streams, was testing
their lengths with LENGTH... I implemented this test into a
preliminary procedure to test before implementing it into my
MAP-NOFAIL proc. and I found, much to my amazement and dismay,
that this worked when I was tracing LENGTH, and did not work
otherwise...here is the output:


------------------------------

---> (query '(unique (job (? x) (computer wizard))))
{
}


---> (trace length)
<PRIMITIVE-PROCEDURE LENGTH>


---> (query '(unique (job (? x) (computer wizard))))
[ Entering (LENGTH ((((? X) (BITDIDDLE BEN))) *DELAYED*)) ]
[ 2 <== (LENGTH ((((? X) (BITDIDDLE BEN))) *DELAYED*)) ]
{
(UNIQUE (JOB (BITDIDDLE BEN) (COMPUTER WIZARD)))
}


---> (untrace length)
<PRIMITIVE-PROCEDURE LENGTH>


---> (query '(unique (job (? x) (computer wizard))))
{
}


---> (tofu)

============================================================
============================================================

And here is the offending code:

(define (uniquely-asserted? assertion environment-stream)
  (let ((app-assert (qeval (asserted-clause assertion)
			   environment-stream)))
    (cond (app-assert
	   (if (= (length app-assert) 2)
	       app-assert
	       (the-empty-stream)))
	  (else (the-empty-stream)))))

(define asserted-clause car)

(put!-prop 'unique 'qeval uniquely-asserted?)

------------------------------


Have I found some kind of bug in this beautiful language? Or am I
incorrect in assuming that tracing a procedure should have no
effect on its performance and evaluation???

This is very curious...

			Thanx,

			  -zzz-
-------

Date: 22 April 1982 17:47-EST
From: Christopher P. Hanson <HANSON at MIT-AI>
Subject: scheme-bug
To: LS.ZZZ at MIT-EECS
cc: BUG-SCHEME at MIT-AI



with the following name changes, the manual is correct:

READC --> READCH
PEEKC --> PEEKCH
CHAR --> CVTA		(convert to ascii)
ASCII --> CVTN		(convert to number)

while ASCII is present and does the same thing as CVTA (don't ask me why),
I would suggest using CVTA instead, since it seems to be the right thing.

thanks,
chris

Date: 22 Apr 1982 1642-EST
From: Michael J. Konopik <LS.ZZZ at MIT-EECS>
Subject: scheme-bug
To: bug-scheme at MIT-EECS

There are MANY things in the user input procedure page of the notes (the
page including such things as READ) that are totally bogus:

READC does not exist
nor does CHAR
ASCII does what CHAR is supposed to do

that is all I have bothered to find so far, I will look through the rest and
report any more...

				Thanks...
				  
				  -Mike-
-------

HANSON@MIT-MC 04/15/82 07:07:37
To: CWH at MIT-MC
CC: (BUG SCHEME) at MIT-MC

the new version of SCHEME on MC contains an improved advice system.
UNADVISE, UNTRACE, and UNBREAK now take multiple arguments, with 0
arguments meaning "all of the advised procedures".  in addition, there
are routines UNADVISE-ENTRY, UNADVISE-EXIT, UNTRACE-ENTRY,
UNTRACE-EXIT, UNBREAK-ENTRY, and UNBREAK-EXIT, which do selective
removal of advisors.

the basic change is that now each function has a SET of advisors for
entry, and another for exit (before there was a single advisor (or
none) for each.  saying UNTRACE, for example, removes only the tracing
advisors, leaving the others intact.  each time the function is
called, the advisors are called sequentially; the actual sequence is
determined by the time at which the advisors are installed -- later
installations are wrapped around earlier ones.

this new system will give you the feature you wanted, namely

(UNTRACE)

will now remove tracing from all traced functions (it will not remove
breakpoints, however, unlike maclisp).

-chris


Date: 15 April 1982 00:07-EST
From: Christopher P. Hanson <HANSON at MIT-AI>
To: GJS at MIT-AI, CWH at MIT-MC
cc: BUG-SCHEME at MIT-AI


my apologies -- the DEFINE bug is mine! i somehow overlooked it when
implementing DEFINE-EXPORT.  one day i shall learn to be careful.

-chris

Date: 12 April 1982 17:33-EST
From: Carl W. Hoffman <CWH at MIT-MC>
To: BUG-SCHEME at MIT-MC
cc: CWH at MIT-MC

In the Scheme installed on MC:

1)   --> (DEFINE (F X) (* 2 X))
     F
     --> F
     <COMPOUND-PROCEDURE &LAMBDA-PROCEDURE>

     This makes TRACE almost totally useless.

2)   PP should not be controlled by the values of *PRINT-BREADTH* and
     *PRINT-DEPTH*.  One often sets these variables to small values when
     debugging circular data structures, but PP should print a procedure
     in its entirety.

3)   UNTRACE of no args should untrace all traced functions.


HANSON@MIT-MC 04/05/82 16:03:31
To: (BUG SCHEME) at MIT-MC

loading STREAM and PRIMES, then typing (FIRST-N-PRIMES 1000) causes a
PDL-OVERFLOW (eventually).  this is strange, and i hope i didn't screw the system.

HANSON@MIT-MC 04/05/82 15:16:26
To: (BUG SCHEME) at MIT-MC

SCHEME version number 140 on mc has undergone a fairly substantial rewrite,
which considerably simplified some of the internal structure of the run-time system,
and added a few new features.

new features implemented are:

(SETUP-READER-HISTORY n)
(SETUP-PRINTER-HISTORY n)
(GET-READER-HISTORY)
(GET-PRINTER-HISTORY)

these functions control the read/print histories, which basically keep track
of all of the terminal-io interactions.  the initial number of cells in each history
is 5 and 10 respectively.  for simplicity, the implementation of %IN/%OUT has
not been changed at all.

file-io is now standard in NEWSYS; however, all file-io has been moved into the
package TRANSPUT-PACKAGE, and the exports include only those functions advertised
in the manual.  this is as opposed to the old newsys, where the environment actually
contained some file-io functions.  the current functions are:

FILE-READ
FILE-TYO
FILE-NEWLINE
FILE-PRINC
FILE-PRIN1
FILE-PRINT
FILE-PRINTC
FILE-OPEN
FILE-CLOSE
TERMINAL-INPUT
TERMINAL-OUTPUT

see the code for details; they work almost exactly like maclisp.

the special form (DEFINE-EXPORT <definition-spec> <package> . <body>) works exactly
like DEFINE, except that the definitions happen in <package> (which is evaluated).

i have not installed the software on EE.  please check it out at your leisure; i have
satisfied myself that it seems ok.

- chris


25-Oct-82 08:08:15-EDT,697;000000000000
Date: Sunday, 24 October 1982  23:07-EDT
From: GJS at MIT-EECS
To:   Clifford Neuman <BCN at MIT-EECS>
Subject: Init.Scm
CC:   bug-scheme at MIT-EECS

    Date: Sunday, 24 October 1982  22:08-EDT
    From: Clifford Neuman <BCN>
    To:   gjs
    Re:   Init.Scm

    Students should be informed of the fact that this file is automatically
    loaded into scheme when it starts up. One student had a problem when 
    he named a file with this name, and found that it had already been 
    loaded when he tried to do it manually.

    Sincerely 

    Cliff

Thanx.  wow! what a potential screw that is.  Probably init files
should have more random names, like scheme_init.scm.
25-Oct-82 14:48:56-EDT,1069;000000000000
Date: Monday, 25 October 1982  14:00-EDT
Sender: HANSON at MIT-OZ
From: HANSON at MIT-MC
To:   Clifford Neuman <BCN at MIT-EECS>
Cc:   bug-scheme at MIT-EECS
Subject: Init.Scm


    Date: Sunday, 24 October 1982  23:07-EDT
    From: GJS at MIT-EECS
    To:   Clifford Neuman <BCN at MIT-EECS>
    cc:   bug-scheme at MIT-EECS
    Re:   Init.Scm

        Date: Sunday, 24 October 1982  22:08-EDT
        From: Clifford Neuman <BCN>
        To:   gjs
        Re:   Init.Scm

        Students should be informed of the fact that this file is automatically
        loaded into scheme when it starts up. One student had a problem when 
        he named a file with this name, and found that it had already been 
        loaded when he tried to do it manually.

        Sincerely 

        Cliff

    Thanx.  wow! what a potential screw that is.  Probably init files
    should have more random names, like scheme_init.scm.

This now looks for "SCHEME.INIT", in keeping with MacLISP ("LISP.INI")
and the LISP Machine ("LISPM.INIT").
-------
25-Oct-82 22:17:19-EDT,493;000000000000
Date: 25 Oct 1982 2209-EDT
From: Mark A. Hamilton <LS.MAH at MIT-EECS>
Subject: Trace, advise,....
To: bug-scheme at MIT-EECS


	The advise type primitives in the version created today (10/25) seem
to be broken.  The functions cause an "Undefined function in UUO call" error,
and leave the user's procedure partially advised.  The problem seems to stem
from somewhere in make-advice-name.  The error is very reproduceable, try
almost anything.

ls.mah
(Mark Hamilton)

-------
26-Oct-82 01:15:17-EDT,360;000000000000
Mail-From: HANSON created at 26-Oct-82 01:11:52
Date: Tuesday, 26 October 1982  01:11-EDT
Sender: HANSON at MIT-OZ
From: HANSON at MIT-MC
To:   LS.MAH at EE
Cc:   BUG-SCHEME at MIT-OZ


The recent bug with tracing behaviour has been fixed.
Sorry about that, I'm finally figuring out how to rebuild
this system without breaking it.

Chris
-------
18-Nov-82 17:09:43-EST,875;000000000000
Date: 18 Nov 1982 1702-EST
From: paul fuqua <LS.PF at MIT-EECS>
Subject: Pretty-Printing
To: bug-scheme at MIT-EECS, bug-6001 at MIT-EECS, bug-scheme at MIT-EECS


     The idea of pretty-printing the output of tracing is appearing worse and 
worse as the semester crawls on.  At first, only the Dataflow people were caught
by the margin problem.  In PS 7 (the evaluator), many problems needed tracing to
locate, and figuring out which procedures would simultaneously produce useful
tracings and not get caught at the right margin was not a trivial process.  Now
try to trace SAVE or RESTORE for homemade RFACT code.  It would be nice to be
able to find out what is really on the stack.  I guess I'll have to do without,
though.  An idea:  how about a variable to determine pretty-printing or no, a la
*print-breadth* ?
                           pf
-------
18-Nov-82 21:08:28-EST,248;000000000000
Date: 18 Nov 1982 2100-EST
From: GJS at MIT-OZ
Subject: Re: Pretty-Printing
To: LS.PF at MIT-EECS
cc: bug-scheme at MIT-EECS, bug-6001 at MIT-EECS, bug-scheme at MIT-EECS
In-Reply-To: Your message of 18-Nov-82 1709-EST

great idea
-------
19-Nov-82 22:31:36-EST,888;000000000000
Date: 19 Nov 1982 2227-EST
From: Ondria Jaffe <LS.LAMB at MIT-EECS>
Subject: The READer needs a shrink!
To: bug-scheme at MIT-EECS

It seems  that the  READer currently  being used  by Scheme  does  not
handle tabs correctly.  My guess is  that it does not convert them  to
spaces.  The  problem  arose when  I  defined a  function  inside  the
brainwashed Lisp driver loop (although  at this writing I haven't  had
the time to see if the outer level of Scheme acts in the same way).  I
used tabs to "pretty-print" the definition, and when the function  was
executed it died--it's  last words were  "Unbound variable: Level  2."
However, when  I  reset Scheme  (etc.)  and typed  the  same  function
replacing the tabs with spaces, the error went away.  Incidentally,  I
was typing the IFACT routine for problem 6 of PS-8.

				Sincerely,

				Ondria G. Jaffe
-------
25-Nov-82 22:22:24-EST,620;000000000000
Date: 25 November 1982 22:19-EST
From: Hal Abelson <HAL at MIT-MC>
To: bug-scheme at MIT-OZ, jmiller at MIT-OZ



I made some changes to the debugger and wrote a new version of

SRC:<SCHEME.DEVELOPMENT>DEBUG.SCM

The major change is that the B and F commands move up and down when
they reach the final reduction at their level, thus making B and F
equivalent to moving up and down "the stack".

I changed the documentation in 20MAN and SCHMAN to reflect my changes.


I don't know how to install this stuff in the regular Scheme on OZ.
Would someone either do this or tell me how? (preferably both)

26-Nov-82 10:20:46-EST,593;000000000000
Mail-From: AAB created at 26-Nov-82 10:20:14
Date: 26 Nov 1982 1020-EST
From: Andrew A. Berlin <AAB at MIT-OZ>
Subject: SCHEME not KEEPING
To: bug-scheme at MIT-OZ


When you first run SCHEME, it does not give you the [keeping]
message.

Possibly related problems:        EMACS buffer not loaded into scheme
                                        after doing an ESC O

                                  After doing a (QUIT), just typing
                                  "SCHEME" to exec gets you a new scheme.

ABOVE REFERS TO OZ SCHEME as of 10:20 a.m., 11/26/82.
-------
26-Nov-82 18:34:17-EST,269;000000000000
Mail-From: GJS created at 26-Nov-82 18:33:40
Date: 26 Nov 1982 1833-EST
From: GJS at MIT-OZ
To: HAL at MIT-MC
cc: bug-scheme at MIT-OZ, jmiller at MIT-OZ
In-Reply-To: Your message of 25-Nov-82 2222-EST

I will install it and show you when you come in.
-------
26-Nov-82 19:54:20-EST,390;000000000000
Mail-From: AAB created at 26-Nov-82 19:54:09
Date: 26 Nov 1982 1954-EST
From: Andrew A. Berlin <AAB at MIT-OZ>
Subject: Problem with Scheme use from SRC:
To: bug-scheme at MIT-OZ
cc: aab at MIT-OZ


If you are connected to an SRC: directory, scheme will not load your

emacs buffer when you do an ESC-O 

Is this a bug or is scheme only supposed to be used from PS: ?
-------
26-Nov-82 20:04:07-EST,425;000000000000
Mail-From: HAL created at 26-Nov-82 20:00:55
Date: 26 Nov 1982 2000-EST
From: HAL at MIT-OZ
To: bug-scheme at MIT-OZ

Changing the debugger uncovered a slight bug in the original debugger,
when it could be put in a state where it calls UNSYNTAX on NIL.

I changed the definition of UNSYNTAX so that (UNSYNTAX NIL) gives NIL
rather than an error.  I did not yet recompile <SCHEME.DEVELOPMENT>SYNTAX.LISP.

-------
28-Nov-82 19:00:28-EST,457;000000000000
Mail-From: HANSON created at 28-Nov-82 18:59:39
Date: Sunday, 28 November 1982  18:59-EST
Sender: HANSON at MIT-OZ
From: HANSON at MIT-MC
To:   Andrew A. Berlin <AAB at MIT-OZ>
Cc:   bug-scheme at MIT-OZ
Subject: Problem with Scheme use from SRC:


Unfortunately, this is a Maclisp bug which probably won't get fixed.
To win, do 

(LISP-EVAL '(SETQ DEFAULTF (MERGEF "SRC:<AAB>" DEFAULTF)))

when you start up SCHEME.  It should then work OK.
29-Nov-82 17:24:22-EST,461;000000000000
Date: 29 Nov 1982 1719-EST
From: Richard Mark Soley <SOLEY at MIT-EECS>
Subject: Bug in equal? function.
To: BUG-SCHEME at MIT-EECS
cc: Df.Jgs at MIT-EECS

If A and B are non-eq? Scheme environments, (equal? A B)
fails with a Maclisp regpdl overflow.  You and I may
understand what's going on, but the average user sure
doesn't, let alone the "REGPDL overflow" message itself.
Perhaps EQUAL? could check for environment-ness?

	-- Richard
-------
29-Nov-82 22:42:15-EST,661;000000000000
Date: 29 Nov 1982 2232-EST
From: Mark A. Hamilton <LS.MAH at MIT-EECS>
Subject: debugger
To: bug-scheme at MIT-EECS
cc: jinx at MIT-OZ


	I went into the R(eturn) mode in the debbugger by mistake, and
(not knowing how to exit) I typed a bunch of ^U's and ^G's.  I assume by
the reaction that these are ignored somehow.  However, when I finally got
back to the lazy-loop, and then went into E(val-in-evironment) mode, the
^G's and ^U's sudenly appeared (and ruined my environment stack).  If there
is a way, these characters should be removed from the input stream, as well
as being ignored while in R(eturn).

Thanx.

Mark (ls.mah)

-------