00100	∂31-JUL-76  1142	FTP:LEFAIVRE at RUTGERS-10	BUGS IN LISP 1.6 COMPILER   
00200	Date: 31 Jul 1976 (Saturday) 1442-Est
00300	From: LEFAIVRE at RUTGERS-10
00400	Subject: BUGS IN LISP 1.6 COMPILER
00500	To:   DIFFIE at SU-AI
00600	
00700	Whit-
00800	
00900	I have no idea who (if anyone) is currently responsible for the LISP
01000	compiler, so I'm sending this to you since your name is on the most
01100	recent manual I've seen.  Would you please pass this on to the appropriate
01200	person if you are not he.
01300	
01400	I'm making a series of improvements to the UCI LISP system when I can
01500	find the time, including an INTERLISP-like prettyprint package, an "octal
01600	point" (Q) to get rid of the BASE/IBASE problem, a generalized CATCH/THROW
01700	facility, and a number of other (mostly minor) changes and additions.  One
01800	change I recently made involved making ERRSETs compile in-line, getting
01900	rid of all of those damn SUBFUNs and their associated special vars.  It
02000	works fine, but while fooling around with the compiler I came across a
02100	bug having to do with special vars being bound by nested internal lambdas.
02200	It is typified by code like the following:
02300	
02400		(DE BUGGY NIL
02500			  ((LAMBDA (L A) (FUNC) L)
02600			   L
02700			   ((LAMBDA (L) (FUNC) (CAR L)) L)))
02800	
02900	In addition to some extra pushes and pops, L is referenced incorrectly
03000	by both lambda exprs.  This is evidently an old bug, and also occurs when
03100	compiling the UCI LISP editor function EDIT*.
03200	
03300	My question: has any further debugging been done on the compiler in the
03400	last few years (i.e., since the UCI group grabbed the compiler)?  In
03500	particular, has this bug been fixed?  If so, I would sure like to get
03600	a list of the fixes.
03700	
03800	Thanks for your help.
03900	
04000	-Rick LeFaivre
04100	
     

00100	∂09-AUG-76  1147	FTP:LEFAIVRE at RUTGERS-10	Debugged version of COMPLR  
00200	Date:  9 Aug 1976 (Monday) 1448-Est
00300	From: LEFAIVRE at RUTGERS-10
00400	Subject: Debugged version of COMPLR
00500	To:   DIFFIE at SU-AI
00600	
00700	Whit-
00800	
00900	I wanted to thank you for the information on the compiler.  I took the
01000	liberty of copying your COMPLR.NOT and found it quite helpful.  It
01100	confirmed many of the things I had figured out by code reading, and
01200	answered several questions I hadn't figured out.  I spent last week
01300	messing around with the code, and as I got more familiar with what's
01400	happening decided to go ahead and fix any problems I found.  I copied
01500	COMPLR[LSP,LSP] from SAIL, and as it seemed to be a somewhat more
01600	recent version than the one that comes with UCI LISP, I decided to work
01700	with it (I had already concluded that I needed something like SPLDLST
01800	anyway).
01900	
02000	The bug which prompted all this was present in both versions of the
02100	compiler.  I discovered that it was somewhat more general than I
02200	had suspected -- it involved having a copy of a special variable
02300	being placed on the PDL without being properly frozen.  This was
02400	done by INTERNALLAMBDA, but could also be brought about via P2SETQ,
02500	as with
02600	
02700		(DE BUG NIL (FUNC (SETQ L 1) (FUNC1) L))
02800	
02900	an (L) gets pushed onto PDL, causing the last L to reference it
03000	instead of the special cell.  I fixed this by insuring that nobody
03100	pushes unfrozen values of special variables on the stack.  I also
03200	made a number of other changes which fix all of the bugs which I
03300	am aware of (i.e., all those in CTEST and CBUGS), in addition to
03400	some of the correct but inefficient code generation (e.g., useless
03500	pushes and pops).  I have one copy of COMPLR from SAIL intact except
03600	for my changes, and another with modifications for UCI LISP which
03700	I will release with my new version.  If you would be interested in
03800	getting it, let me know (note, however, that I'm leaving for a
03900	vacation on Nantucket tomorrow and won't be back till August 21).
04000	
04100	Although all of the bugs seem to be out of the compiler, there is
04200	still something which is confusing me.  The old compiler I have
04300	has a function GUARDLOCS which is called from P2COND and P2PROG,
04400	apparently to force any values waiting to be loaded to get a place
04500	on the PDL before the COND or PROG is entered so that they can
04600	shrink the stack without worrying about them.  In the newer compiler
04700	GUARDLOCS has disappeared, with P2COND providing this service only
04800	for those vars which will be set inside the COND.  It seems to me
04900	that this is not necessary -- CLEAR1 will make sure that valuable
05000	stuff gets placed on the stack before the COND is entered.  If it
05100	is necessary for some reason, why doesn't PROG also have to do the
05200	same thing?  I was tempted to just get rid of this all together, but
05300	it was one of the few things I didn't really understand and was thus
05400	hesitant to fool with it.  Are you still close enough to the compiler
05500	to fill me in on what's going on here?
05600	
05700	-Rick
05800	
     

00100	∂09-AUG-76  1335	FTP:LEFAIVRE at RUTGERS-10	WHAT HAPPENED TO COPT? 
00200	Date:  9 Aug 1976 (Monday) 1636-Est
00300	From: LEFAIVRE at RUTGERS-10
00400	Subject: WHAT HAPPENED TO COPT?
00500	To:   DIFFIE at SU-AI
00600	
00700	Another quick question.  When comparing some LAP code produced by the
00800	new compiler I got from SAIL with that produced by the seemingly older
00900	UCI compiler, I noticed that the new compiler computed a CDR when it
01000	really didn't have to, whereas the old compiler didn't.  Upon investigation
01100	I found that the call to COPT in CSTEP had been removed (in fact, COPT is
01200	not used anymore).  Do you know the reason for this?  Is there any reason
01300	I shouldn't reinstate its use?
01400	
01500	-Rick
01600	
01700	∂22-AUG-76  0857	FTP:LEFAIVRE at RUTGERS-10	MORE ON THE LISP COMPILER   
01800	Date: 22 Aug 1976 (Sunday) 1155-Est
01900	From: LEFAIVRE at RUTGERS-10
02000	Subject: MORE ON THE LISP COMPILER
02100	To:   DIFFIE at SU-AI
02200	
02300	Whit-
02400	
02500	I'm now back from vacation and trying to get back in the swing and finish
02600	up my fixes to the compiler in the next few days.  Did you get my letter
02700	of August 9?  I'm still unclear as to the removal of GUARDLOCS in the version
02800	of COMPLR in [LSP,LSP].
02900	
03000	I also noticed that the new compiler recomputed a CAR at a time when it had
03100	it sitting in a register -- the old compiler knew that it was already there.
03200	I traced this down to the fact that the call to COPT in CSTEP was removed for
03300	some reason (in fact, COPT is not even used in the new version).  Do you
03400	know why this was done?  Would some obscure bug reoccur if I reinserted it?
03500	
03600	-Rick
03700