LISP TROUBLE REPORT LIST

TR numbers are assigned sequentially by report date.
Unresolved TR's are indicated by a ? preceding the TR number in column 1.
Resolved TR's are indicated by the ? being replaced by a period.


		Alpha test trouble reports (pre version 1)


.1	6-30-83	Kelley Miller	(closed 7-8-83, unable to reproduce problem)
Using ATOMCAR/ATOMCDR, subtring of atoms, ob-list got null symbols on it, and
REMOB wouldn't work.  Had to get out and reenter. 
[See TR#6 for possible explanation of how this might have occurred.]

.2	6-30-83	GPW	(fixed 6-30-83)
PPD (from PP.LSP) bombed on compiled functions.

.3	6-30-83 GPW	(fixed 7-8-83)
Various problems in ATOMCAR & ATOMCDR discovered as a result of investigating 
TR #1.

.4	7-5-83 Kelly Miller	(fixed 7-12-83)
MEMBER doesn't work for numeric arguments.

.5	7-11-83 GPW	(fixed 7-12-83)
APPLY fails to apply functions that are atomic.

.6	7-12-83 GPW	(work around developed 7-12-83; fixed 7-14-83)
If a builtin form is ever redefined, it's property list will be destroyed by
any garbage collection.
This problem mey be eliminated by using REMOB to delete the form from the
ob-list before [re-]defining it.

.7	7-14-83 GPW	(fixed 7-14-83)
When logging input and output (via LOGIO), and an error occurs,
part of the error message is not logged with a leading comment character.

.8	7-21-83 GPW	(fixed 7-21-83)
DE, DF, DM do not enforce the requirement that the argument list of 
the form being defined must be a list, causing problems later at 
evaluation time.

.9	7-28-83 GPW	(fixed 7-28-83)
READCH returns !*EOL instead of !$EOL!$, as specified for Standard LISP.

			Beta Test TR's
			 (Version 1)

.1.1	8-10-83 GPW	(fixed 8-10-83)
The LOOP macro delivered with Version 1 would not operate correctly when
a WHILE clause was included.

.1.2	8-15-83 GPW	(fixed 8-16-83)
The lexical analyzer had problems with a few special cases where there was a 
quoted string on an input line, and the line was terminated with an identifier 
or a number.  There were also some similar problems with input lines 
containing ASCII NUL (hex zero) characters.  The problem typically was 
experienced only with input that had been squeezed to eliminate white space to 
speed input to LISP.

.1.3	8-15-83 GPW	(fixed 8-16-83)
The READ function does not always pay proper attention to the character 
classes assigned by the LEX function (or by default) for characters of class 
CHRCLWHSPACE, CHRCLNULL, or CHRCLTERMN when reading standard list notation.
Certain restrictions that are intrinsic to the implementation on the placement
of comments using CHRCLCOMBEG and CHRCLCOMEND class characters within standard
list notation were recognized and documented in the help file under the 
documentation for character class CHRCLCOMBEG under the function LEX.

.1.4	8-16-83 GPW	(fixed 8-16-83)
TERPRI has always been defined in a slightly non-standard manner, specifically 
such that it printed anything that was in the output buffer, but would not 
print a blank line.  This is now fixed according to the standard definition.

.1.5	8-17-83 GPW	(fixed 8-18-83)
INTERN interned the string that corresponded to the print name of the atom 
most recently read or INTERNed when its argument was a string or an uninterned 
atom.  It also did not intern an uninterned atoms properties when that atom 
was interned.

.1.6	8-18-83 GPW	(fixed 8-18-83)
READ/PRIN1/PRIN2 -- When the output buffer is partially filled and a READ is 
performed, the output buffer is not printed as a prompt.  The fix doesn't 
completely fix this, but at least the output buffer is flushed in this 
situation.

?1.7	9-9-83	GPW
It's not clear yet, but for some reason, the definition of E in PP.LSP is 
causing the PNAME property of E to be placed on its property list twice.
(This is also true of the version built into the interpreter.)

.1.8	9-14-83	GPW	(fixed 9-14-83)
The help file was found to be in error in the description of the returned 
value for the functions ASSOC and SASSOC.

.1.9	9-14-83	GPW	(fixed 9-14-83)
The functions ASSOC and SASSOC were searching the association lists using only 
an EQ comparison, instead of an EQUAL comparison.

			Trouble Reports
			 (Version 2)

.2.1	2-17-84 Ahmad Alavi (fixed 3-1-84)
When the R5 stack overflows, an access violation occurs.
A more descriptive error message would be more useful.

.2.2	5-4-84 GPW (fixed 5-7-84)
MAPCAR passes successive CDR segments of the list.

.2.3	5-7-84 GPW (no fix)
APPLY does not generate an error if an EXPR is called with more arguments than
the function is defined to accept.
Although this behavior is less than desirable, it will remain for the present
time.  This is because this implementation makes possible FEXPRs with zero, one
or two formal parameters, regardless of the number of arguments.

.2.4	5-7-84 GPW (fixed 5-8-84)
RPLACA and RPLACD do not verify that their first argument is a list, resulting
in access violations and other improper errors.

.2.5	5-7-84 GPW (fixed 5-8-84)
PUT does not verify that its property name argument is an id.

.2.6	5-7-84 GPW (fixed 5-8-84)
REMOB returns NIL instead of the id removed from the oblist, as called for in
the LISP Standard Report. 

.2.7	5-7-84 GPW (fixed 5-8-84)
REMPROP returns NIL instead of the property value removed from the id's
property list, as called for in the LISP Standard Report. 

			Trouble Reports
			 (Version 3)

.3.1	6-21-84 GPW (fixed 6-21-84)
SASSOC did not evaluate the default function argument when the item was not
found on the alist.

.3.2	6-21-84 GPW (fixed 6-21-84)
ASSOC and SASSOC did not correctly detect an improperly formed alist as 
indicated in the Report.

.3.3	6-22-84 GPW (fixed 6-22-84)
EVAL had a potential infinite loop in an extension to the standard.
The extension permitted EVAL to check the alist for an association for an atom 
which was the CAR of a list being evaluated after GETD was used to sucessfully 
obtain a definition for this atom and the definition was not one which EVAL 
knew how to process.  This alist search is now permitted only when GETD fails 
to find a definition for the atom.

.3.4	6-22-84 GPW (fixed 6-22-84)
The reference manual indicated that the TIMES function was implemented 
according to the standard, which was incorrect.  TIMES was actually the 
standard TIMES2 function.  Both of these functions are now implemented.

.3.5	6-26-84 GPW (fixed 6-26-84)
The PPD function produced improper output for MACROs.

.3.6	6-26-84 GPW (fixed 6-26-84)
TRACE clobbered itself when handed a null argument list.

.3.7	7-9-84 GPW (fixed 7-9-84)
NUMBERP aborts with an access violation if called with a null argument.

?3.8	7-10-84 GPW
The pretty printer routines do not properly handle dotted pairs, or lists 
whose last element's CDR is non-NULL.  This is not considered a severe problem 
at this time, and therefore a fix does not now seem worth the effort.

			Trouble Reports
			 (Version 4)

.4.1	7-26-84 GPW (fixed 7-26-84)
Garbage collection improperly collected vector values which are not otherwise
referenced.  Garbage collection could also conceivably incorrectly collect 
things that were allocated in pool expansion areas (created using GROWMEM).

.4.2	7-27-84 GPW (fixed 7-27-84)
Many of the arithmetic related functions did not have adequate type checking.
This showed up when these functions were inadvertantly passed non-numeric data 
as results that were less than reputable.  The error checking has been 
improved in the following functions:
	ABS	ADD1	DIFFERENCE	EQN	FIX	FLOAT	GREATERP
	LESSP	MAX	MAX2	MIN	MIN2	MINUS	PLUS	PLUS2
	QUOTIENT	REMAINDER	SUB1	TIMES	TIMES2	ZEROP

.4.3	8-1-84 GPW (fixed 8-1-84)
The BREAK function returned garbage.

.4.4	8-1-84 GPW (fixed 8-1-84)
The EQUAL function incorrectly returned NIL for numeric comparisons under
extremely rare circumstances.

.4.5	8-6-84 GPW (fixed 8-6-84)
Math functions requiring two arguments would get VAX Access Violations when
passed a null argument.

.4.6	8-8-84 GPW (fixed 8-8-84)
SUBST did not correctly perform it's comparison for equality in the case of
numeric values.

.4.7	8-8-84 GPW (fixed 8-8-84)
FLOAT did not work correctly.

.4.8	8-8-84 GPW (fixed 8-8-84)
Garbage collection had a bug that caused it to corrupt the node pool under
several unusual circumstances.

.4.9	8-14-84 GPW (fixed 8-14-84)
The BREAK function would get an access violation if it read a NIL or the user
entered a readmacro that resulted in NIL.

			Trouble Reports
			 (Version 5)

.5.1	8-21-84 GPW (fixed 8-21-84)
READCH and COMPRESS/IMPLODE incorrectly interned their results when the result
was already on the oblist.

.5.2	8-23-84 GPW (fixed 8-24-84)
There was an obscure error in the garbage collection algorithm which caused it 
to collect things which were not garbage in extremely rare circumstances.