search

Interlisp / medley

The main repo for the Medley Interlisp project. Wiki, Issues are here. Other repositories include maiko (the VM implementation) and Interlisp.github.io (web site sources)
https://Interlisp.org
MIT License
370 stars 19 forks source link

Problems with ACE LispUsers MENU code -- works in MEDLEY 1.0 #525

Open masinter opened 2 years ago

masinter commented 2 years ago

@mdenber reports

Still trying to figure out why ACE doesn't work. I tried it on a Medley 1 ldex and it works fine there. I think something changed in ADDMENU between Medley 1 and 3.

How can I get access to Medley 1?

masinter commented 1 year ago

When you start ACE up, it writes on the background rather than in the window. image

you have to be careful debugging things that might be used y the debugger. if you (NREAK ADDMENU) you can get a stack overflow. (BREAK (ADDMENU IN AVE)) might help localize this problem.

masinter commented 1 year ago

this is a bug in BITBLT(!). There's the image of the menu in SS and a window DST. if I make a new window of the same width and height 568v450 and add USERDATA image ....

nbriggs commented 1 year ago

@masinter - I think part of that message got lost. If it's a basic BITBLT operation, that would be implemented in the Maiko C code, and I would be quite surprised that this was the first time such a problem had shown up.

pictographer commented 1 year ago

When I run it, I see a variety of bad behaviors. Sometimes it draws the menu and sometimes it doesn't. ACE caches several values, so running it the second and subsequent time don't execute the exact same code path as the first time. I've observed it writing to the screen background and also breaking in ACE.QUICKDRAW&UPD in DSPCLIPPINGREGION with \ILLEGAL.ARG NIL when attempting to loop an animation, but after the first instance of this, it's not clear that we're seeing anything other than side-effects of a previous bug.

Michele31415 commented 1 year ago

On 3/26/2023 4:19 PM, John D. Corbett wrote:

When I run it, I see a variety of bad behaviors. Sometimes it draws the menu and sometimes it doesn't. ACE caches several values, so running it the second and subsequent time don't execute the exact same code path as the first time. I've observed it writing to the screen background and also breaking in ACE.QUICKDRAW&UPD in DSPCLIPPINGREGION with \ILLEGAL.ARG NIL when attempting to loop an animation, but after the first instance of this, it's not clear that we're seeing anything other than side-effects of a previous bug.

That's all true.  I've witnessed it myself.  I must though humbly reiterate that it all worked fine in Medley 1.0.  I don't think there's any code in there that explicitly draws on the screen bitmap.

            - Michele

pictographer commented 1 year ago

Are there any screen images of what the ACE many window is supposed to look like?

ADDMENU clobbers various window properties to make the menu work. ACE clobbers them back. Wondering if somewhere a display stream is now somehow defaulting to the the screen background because some function that used to return the application window now returns NIL or something like that.

Understood that it works in Medley 1.0 and whatever version of Interlisp-D it was developed in.

masinter commented 1 year ago

I fiddled around a bit. I made some dribble logs of log.txt log2.txt it looks like the clipping is offset from where it should be, and so the BITBLT writes outside the menu-window's boundaries. i ran new ace in loadups/lisp.venuesysout and old ace in current.

Michele31415 commented 1 year ago

On 3/26/2023 8:20 PM, John D. Corbett wrote:

Are there any screen images of what the ACE many window is supposed to look like?

Like this:

This is in a Medley 1.0 sysout with a MAKESYSDATE of 28-Aug-91 running with the Solaris SPARC ldex I found on the PARC server, file size 2278016.

            - Michele

pictographer commented 1 year ago

Was trying to familiarize myself with ADDMENU and all the rest. Recreating the initial appearance is easy and doesn't show the bug. I added a few menu items for convenience while playing around. ACE_FAKE ACE_FAKE_CODE

The ACE code calls ADDMENU with the location (0 . 0) but I'm not sure how this could be right, (actually not right but lower left).

pictographer commented 1 year ago

Michele, the image didn't appear in your comment.

Michele31415 commented 1 year ago

That's odd. How about this:

image

pictographer commented 1 year ago

@Michele31415, now I see it. So, (0 . 0) is correct. Maybe the buggy x-coordinate the menu gets rendered at is related to the text starting location for the directory prompt.

I wonder if there's a reason for shading the title bar. I don't recall that being a common thing.

pictographer commented 1 year ago

When I change the ADDMENU location to (0 . 0), and change the window border size to 1, the menu is drawn exactly as in Michele's screenshot.

nbriggs commented 1 year ago

It calls BKBITBLT to put up the menu image --

image

and it looks as though it's trying to call it with the correct location (0 . 0) but we'll need to look at what BKBITBLT should be doing.

pictographer commented 1 year ago

I discovered a way to recreate the problem. If we call (DSPXOFFSET X STREAM) or (DSPYOFFSET Y STREAM) the menu renders outside the window and when the window is moved, it redisplays offset by the change in position of the window.

The ACE code does manipulate the window's clipping region despite the caution in the IRM.

nbriggs commented 1 year ago

The code in \BITBLT.DISPLAY does take into account the offsets --

(SETQ DESTINATIONLEFT (\DSPTRANSFORMX DESTINATIONLEFT DESTDD))

and likewise for the bottom, so if under Medley 1.0 the offsets were ignored, but then that "bug" was fixed, we'd see this change. I can't find a copy of LLDISPLAY from Medley 1.0 to check.

Michele31415 commented 1 year ago

@Michele31415,

I wonder if there's a reason for shading the title bar. I don't recall that being a common thing.

I hadn't even noticed that. I checked a screenshot of Lisp running on the D0 and it's not that way there. It's also just a solid line in Medley 3.5. It's a mystery.

masinter commented 1 year ago

I found lots of versions of LLDISPLAY, removing duplicates


(FILECREATED "13-Jun-90 13:28:33" {DSK}<home>ringneck>takeshi>M1.2>ARFIX>LLDISPLAY.;2 159403  ./history/1990s/1993-envos/extra-lispcore-sources/LLDISPLAY: 
(FILECREATED "26-Apr-91 17:00:45" |{PELE:MV:ENVOS}<LISPCORE>LIBRARY>LLDISPLAY.;1| 161932  ./history/1990s/1993-envos/LISPCORE/library/LLDISPLAY:
(FILECREATED "28-Jan-93 17:38:44" {DSK}<python>lde>lispcore>sources>LLDISPLAY.;2 249157  ./history/1990s/1993-envos/LISPCORE/sources/LLDISPLAY:
(FILECREATED "18-Apr-94 00:20:42" {DSK}<users>nilsson>mnw>LLDISPLAY.;7 267646  ./history/202011-lisp/sources/LLDISPLAY:

(FILECREATED " 9-Jul-2022 12:08:02"  ./medley/sources/LLDISPLAY:

I think BKBITBLT did it in the (STORAGE) (L:ROOM) with a Clipping Region

masinter commented 1 year ago

here are 5 versions of LLDISPLAY picked out from the history repo (mainly) lldisplays.zip

pamoroso commented 1 year ago

@pictographer wrote:

Are there any screen images of what the ACE many window is supposed to look like?

The ACE documentation contains a screenshot of the control window:

ace-control-window

and the main menu:

ace-menu

pamoroso commented 1 month ago

As discussed in the Jul 31, 2024 external meeting, to isolate the issue I'll test ACE on Medley for DOS which I managed to run under DOSBox-X as reported in #1790. It has a Medley 2 sysout dated Sep 1992.

pamoroso commented 1 month ago

To run ACE on Medley for DOS I downloaded the compiled files from the Medley repo, renamed them from .LCOM to .LCO, and copied them to C:\MEDLEY\ACE. Then from Medley for DOS I changed to the directory medley>ace, evaluated (il:filesload ace), and got this error:

Bad compiled function
ACE

medley-dos-ace-error

It's probably better to build ACE from source on DOS. Can I just FILE-COMPILE every source file? Any required compilation order?

One of the ACE source files, ACEIMAGEOBJ, has a name longer than the 8+3 DOS format (and there's no compiled version). How should I handle it?

masinter commented 1 month ago

I'm not sure of the timeline of introducing new features in Maiko, but there were incompatible changes between Medley 1.0 and Medley 3.5 (which we started with in current Medley/Maiko):

but in any case Medley 2.0 compiled files and Medley 3.5 current repo compiled files are incompatible. You can't use current medley LCOMs or DFASLsin Medley 2.0. You can use IL:BCOMPL or CL:COMPILE-FILE on each file.

pamoroso commented 1 month ago

I made some progress with compiling ACE.

I downloaded the source files and renamed ACEIMAGEOBJ to ACEIMGOB to fit the DOS format, then compiled them with CL:COMPILE-FILE. The following files compiled with no errors:

The compilation of ACE-EDIT broke into the debugger with the error The form 16, appearing in a functional context, is neither a symbol nor a LAMBDA-form:

ace-edit-compile

Compiling ACE-MAIN broke into the debugger with the same error The form 16, appearing in a functional context, is neither a symbol nor a LAMBDA-form:

ace-main-compile

nbriggs commented 1 month ago

I'd BCOMPL them rather than use the CommonLisp compiler. Also, when you get a break like that, don't just stop... use the backtrace, look back up the stack, look at the form it was compiling, where did it come from, does it make sense?

pamoroso commented 1 month ago

I compiled each ACE file separately with BCOMPL and for all of them the build completed with no errors. Then I loaded the compiled ACE file with (il:filesload ace) and evaluated (il:ace). ACE opened the main window but the program broke into the debugger with the error:

In IL:OLDFAULT1:
IL:DON\'T is an undefined function

ace-start-error

The backtrace:

17: bt
IL:OLDFAULT1
IL:NEWFAULT1
IL:FAULTAPPLY
IL:DON\'T
IL:DOUSERFNS2
SI::*UNWIND-PROTECT*
IL:REDISPLAYW
IL:MOVEW
IL:ACE
APPLY
XCL::NOHOOK
(IL:ACE)
EVALHOOK
XCL::UNDOHOOK
XCL::|interpret-UNDOABLY|
(UNDOABLY (IL:ACE))
IL:EVAL-INPUT
IL:DO-EVENT
XCL::EXECA0001A0002
XCL::EXECA0001
EXEC
T

Some more debugger commands:

21: ?=
  FAULTX = IL:DON\'T
  FAULTARGS = (#<IL:WINDOW @ 65,150234> (299 0 162 ...) NIL ...)
  FAULTAPPLYFLG = T
22: pb faultx
bindings for FAULTX: 
 @ TOP : IL:NOBIND
23: pb faultargs
bindings for FAULTARGS: 
 @ TOP : IL:NOBIND
24: pb faultapplyflg
bindings for FAULTAPPLYFLG: 
 @ TOP : IL:NOBIND
25: btv
   IL:FAULTX IL:DON\'T
   IL:FAULTARGS (#<IL:WINDOW @ 65,150234> (299 0 162 99)
 NIL NIL)
   IL:FAULTAPPLYFLG T
IL:OLDFAULT1
IL:NEWFAULT1
   IL:FAULTFN IL:DON\'T
   IL:FAULTARGS (#<IL:WINDOW @ 65,150234> (299 0 162 99)
 NIL NIL)
   IL:DEF NIL
   IL:%LEXICAL-ENVIRONMENT% NIL
   IL:TRAN NIL
   IL:TRANFN NIL
IL:FAULTAPPLY
IL:DON\'T
   IL:FNLST IL:DON\'T
   IL:WINDOW #<IL:WINDOW @ 65,150234>
   IL:ARG1 (299 0 162 99)
   IL:ARG2 NIL
   IL:ARG3 NIL
IL:DOUSERFNS2
   SI::*CLEANUP-FORMS* SI::RESETUNWIND
SI::*UNWIND-PROTECT*
   IL:WINDOW #<IL:WINDOW @ 65,150234>
   IL:REGION (299 0 162 99)
   IL:ALWAYSFLG NIL
   IL:DSP #<Output Display Stream/56,7700>
   IL:REPAINTFN IL:DON\'T
   IL:CLIPREG (0 0 461 125)
   IL:LISPXHIST ((#) (12 "" . "> ") 
"<not yet evaluated>" NIL IL:SIDE (1 #))
   SI::*RESETFORMS* ((# NIL) (# NIL) (# NIL))
   IL:RESETSTATE NIL
IL:REDISPLAYW
   IL:WINDOW #<IL:WINDOW @ 65,150234>
   IL:|POSorX| NIL
   IL:Y NIL
   IL:OLDREGION (500 500 463 139)
   IL:USERMOVEFN NIL
   IL:OPEN? T
   IL:OLDSCREEN #<IL:SCREEN @ 65,151740>
   IL:POS (233 . 100)
   IL:NEWREGION (233 100 463 139)
   IL:OLDLEFT 500
   IL:OLDBOTTOM 500
   IL:OLDWIDTH 463
   IL:OLDHEIGHT 139
   IL:OLDCLIPREGION (0 0 299 99)
   IL:LFT NIL
   IL:BTM NIL
   IL:REG NIL
   IL:FN NIL
   IL:NEWCLIPPINGREGION (0 0 461 125)
   IL:NCL 0
   IL:OCL 0
   IL:NCB 0
   IL:OCB 0
   IL:OCR 299
   IL:NCR 461
   IL:OCW 299
   IL:NCW 461
   IL:OCH 99
   IL:NCH 125
   IL:OCT NIL
   IL:NCT NIL
IL:MOVEW
   SEQUENCE NIL
   IL:WINDOW NIL
   IL:POSITION NIL
   IL:APPLICATION NIL
   IL:FONT #<IL:FONTDESCRIPTOR @ 74,73464>
   IL:TEMP.REGION NIL
IL:ACE
APPLY
   *EVALHOOK* NIL
XCL::NOHOOK
   *APPLYHOOK* NIL
(IL:ACE)
   *EVALHOOK* XCL::UNDOHOOK
   CL::*SKIP-EVALHOOK* NIL
   *APPLYHOOK* XCL::NOHOOK
   CL::*SKIP-APPLYHOOK* NIL
EVALHOOK
   *APPLYHOOK* NIL
XCL::UNDOHOOK
XCL::|interpret-UNDOABLY|
(UNDOABLY (IL:ACE))
IL:EVAL-INPUT
   IL:RETRYFLAG NIL
   IL:HELPCLOCK 527559884
IL:DO-EVENT
   SI::*DUMMY-FOR-CATCH* T
   SI::*CATCH-RETURN-FROM* (#)
   IL:LISPXHIST ((#) (12 "" . "> ") 
"<not yet evaluated>" NIL IL:SIDE (1 #))
   IL:HELPCLOCK 0
XCL::EXECA0001A0002
   *CURRENT-EVENT* ((#) (12 "" . "> ") 
"<not yet evaluated>" NIL IL:SIDE (1 #))
   SI::NLSETQ-VALUE NIL
   IL:*PROCEED-CASES* (#)
   SI::*NLSETQFLAG* NIL
XCL::EXECA0001
IL:\\PROGV
   XCL::TOP-LEVEL-P T
   XCL::WINDOW #<IL:WINDOW @ 65,150640>
   XCL::TITLE-SUPPLIED NIL
   XCL::TITLE NIL
   IL:*THIS-EXEC-COMMANDS* (#<Hash-Table @ 70,37614>)
   XCL::ENVIRONMENT NIL
   XCL::PROMPT NIL
   XCL::FN IL:EVAL-INPUT
   XCL::PROFILE "XCL"
   IL:*EXEC-ID* ""
   XCL::PROFILE-CACHE (XCL::*PROFILE-NAME* "XCL" 
*EVAL-FUNCTION* EVAL *PACKAGE* #<Package XCL-USER> 
*READTABLE* #<ReadTable XCL/74,146670> *EXEC-PROMPT* 
"> " ...)
EXEC
IL:\\PROC.REPEATEDLYEVALQT
   IL:*FORM* (IL:\\PROC.REPEATEDLYEVALQT)
   IL:*ARGVAL* NIL
   IL:*TAIL* NIL
   IL:*FN* IL:\\PROC.REPEATEDLYEVALQT
IL:\\EVALFORM
   IL:\#FORM# (IL:\\PROC.REPEATEDLYEVALQT)
   IL:*CURRENT-PROCESS* #<Process EXEC/74,107300>
   IL:HELPFLAG IL:BREAK!
   IL:\\CURRENTDISPLAYLINE 0
   IL:\\#DISPLAYLINES 25
   IL:\\LINEBUF.OFD #<IO Linebuffer Stream/71,106430>
   *READTABLE* #<ReadTable INTERLISP/74,146726>
   IL:\\PRIMTERMTABLE #<IL:TERMTABLEP @ 74,141740>
   IL:\\PRIMTERMSA #<IL:CHARTABLE @ 74,142000>
   IL:|TtyDisplayStream| 
#<Output Display Stream/71,106520>
   SI::*RESETFORMS* NIL
   IL:\\INTERRUPTABLE T
   IL:\\TTYWINDOW NIL
   IL:READBUF NIL
   IL:\\TERM.OFD #<Output Display Stream/71,106610>
   *STANDARD-OUTPUT* #<Output Display Stream/71,106610>
   *STANDARD-INPUT* #<IO Linebuffer Stream/71,106430>
IL:\\MAKE.PROCESS0
T
nbriggs commented 1 month ago

It's putting a DON'T atom on the ACE.CONTROL.WINDOW's RESHAPEFN, REPAINTFN, and in some cases, the MOVEFN properties - this is standard, see the IRM (DINFO) entry for SHAPEW. Not sure how it got to the point of trying to call DON'T.

nbriggs commented 1 month ago

You could exit the break with (RETFROM 'DON\'T NIL) and see what happens after that.

nbriggs commented 1 month ago

BTW - I would definitely not compile or run this from an XCL exec - it's Interlisp so use an Interlisp exec.

pamoroso commented 1 month ago

From an Interlisp Exec I recompiled the ACE sources with BCOMPL, loaded the ACE file with (FILESLOAD ACE), evaluated (ACE), and got the same error. I executed (RETFROM 'DON\'T NIL) from the debugger and got the error:

Illegal stack arg: DON\'T

ace-dbg-error

The context of the error:

2/8_: BT
OLDFAULT1
NEWFAULT1
FAULTAPPLY
DON'T
DOUSERFNS2
SI::*UNWIND-PROTECT*
REDISPLAYW
MOVEW
ACE
FAULTEVAL
EVAL
EVAL-INPUT
DO-EVENT
XCL::EXECA0001A0002
XCL::EXECA0001
EXEC
T

2/9_: 
2/10_: ?=
  FAULTX = DON'T
  FAULTARGS = ({WINDOW}#65,150000 (299 0 --) NIL --)
  FAULTAPPLYFLG = T
2/11_: PB FAULTX
bindings for FAULTX: 
 @ OLDFAULT1 : DON'T
 @ TOP : NOBIND
2/12_: PB FAULTARGS
bindings for FAULTARGS: 
 @ OLDFAULT1 : ({WINDOW}#65,150000 (299 0 --) NIL --)
 @ FAULTAPPLY : ({WINDOW}#65,150000 (299 0 --) NIL --)
 @ TOP : NOBIND
2/13_: PB FAULTAPPLYFLG
bindings for FAULTAPPLYFLG: 
 @ OLDFAULT1 : T
 @ TOP : NOBIND
2/14_: BTV
   FAULTX DON'T
   FAULTARGS ({WINDOW}#65,150000 (299 0 162 99) NIL NIL)
   FAULTAPPLYFLG T
OLDFAULT1
NEWFAULT1
   FAULTFN DON'T
   FAULTARGS ({WINDOW}#65,150000 (299 0 162 99) NIL NIL)
   DEF NIL
   %%LEXICAL-ENVIRONMENT%% NIL
   TRAN NIL
   TRANFN NIL
FAULTAPPLY
DON'T
   FNLST DON'T
   WINDOW {WINDOW}#65,150000
   ARG1 (299 0 162 99)
   ARG2 NIL
   ARG3 NIL
DOUSERFNS2
   SI::*CLEANUP-FORMS* SI::RESETUNWIND
SI::*UNWIND-PROTECT*
   WINDOW {WINDOW}#65,150000
   REGION (299 0 162 99)
   ALWAYSFLG NIL
   DSP #<Output Display Stream/71,106250>
   REPAINTFN DON'T
   CLIPREG (0 0 461 125)
   LISPXHIST ((&) (5 "2" . "_ ") "<not yet evaluated>" 
NIL SIDE (1 &))
   SI::*RESETFORMS* ((& NIL) (& NIL) (& NIL))
   RESETSTATE NIL
REDISPLAYW
   WINDOW {WINDOW}#65,150000
   POSorX NIL
   Y NIL
   OLDREGION (500 500 463 139)
   USERMOVEFN NIL
   OPEN? T
   OLDSCREEN {SCREEN}#65,151740
   POS (301 . 368)
   NEWREGION (301 368 463 139)
   OLDLEFT 500
   OLDBOTTOM 500
   OLDWIDTH 463
   OLDHEIGHT 139
   OLDCLIPREGION (0 0 299 99)
   LFT NIL
   BTM NIL
   REG NIL
   FN NIL
   NEWCLIPPINGREGION (0 0 461 125)
   NCL 0
   OCL 0
   NCB 0
   OCB 0
   OCR 299
   NCR 461
   OCW 299
   NCW 461
   OCH 99
   NCH 125
   OCT NIL
   NCT NIL
MOVEW
   SEQUENCE NIL
   WINDOW NIL
   POSITION NIL
   APPLICATION NIL
   FONT {FONTDESCRIPTOR}#74,73464
   TEMP.REGION NIL
ACE
   *FORM* (ACE)
   *ARGVAL* NIL
   *TAIL* NIL
   *FN* ACE
\EVALFORM
FAULTEVAL
   *FORM* (UNDOABLY (ACE))
\EVALFORM
   \INTERNAL NIL
EVAL
EVAL-INPUT
   RETRYFLAG NIL
   HELPCLOCK 601376304
DO-EVENT
   SI::*DUMMY-FOR-CATCH* T
   SI::*CATCH-RETURN-FROM* (&)
   LISPXHIST ((&) (5 "2" . "_ ") "<not yet evaluated>" 
NIL SIDE (1 &))
   HELPCLOCK 0
XCL::EXECA0001A0002
   *CURRENT-EVENT* ((&) (5 "2" . "_ ") 
"<not yet evaluated>" NIL SIDE (1 &))
   SI::NLSETQ-VALUE NIL
   *PROCEED-CASES* (&)
   SI::*NLSETQFLAG* NIL
XCL::EXECA0001
\PROGV
   XCL::TOP-LEVEL-P T
   XCL::WINDOW {WINDOW}#65,150150
   XCL::TITLE-SUPPLIED NIL
   XCL::TITLE NIL
   *THIS-EXEC-COMMANDS* (#<Hash-Table @ 70,37614>)
   XCL::ENVIRONMENT NIL
   XCL::PROMPT NIL
   XCL::FN EVAL-INPUT
   XCL::PROFILE "INTERLISP"
   *EXEC-ID* "2"
   XCL::PROFILE-CACHE NIL
EXEC
   *FORM* (EXEC :TOP-LEVEL-P T :PROFILE (QUOTE 
"INTERLISP") :ID (QUOTE NIL))
   *ARGVAL* NIL
   *TAIL* NIL
   *FN* EXEC
\EVALFORM
   *TAIL* ((EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
PROGN
   *FORM* (PROGN (TTYDISPLAYSTREAM &) (PROCESSPROP & & &
) (EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
\EVALFORM
   %#FORM# (PROGN (TTYDISPLAYSTREAM &) (PROCESSPROP & & &) (EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
   *CURRENT-PROCESS* #<Process EXEC#2/74,107000>
   HELPFLAG BREAK!
   \CURRENTDISPLAYLINE 0
   \#DISPLAYLINES 24
   \LINEBUF.OFD #<IO Linebuffer Stream/74,23070>
   *READTABLE* #<ReadTable INTERLISP/74,146726>
   \PRIMTERMTABLE {TERMTABLEP}#74,141740
   \PRIMTERMSA {CHARTABLE}#74,142000
   TtyDisplayStream #<Output Display Stream/71,106520>
   SI::*RESETFORMS* NIL
   \INTERRUPTABLE T
   \TTYWINDOW NIL
   READBUF NIL
   \TERM.OFD #<Output Display Stream/71,106340>
   *STANDARD-OUTPUT* #<Output Display Stream/71,106340>
   *STANDARD-INPUT* #<IO Linebuffer Stream/74,23070>
\MAKE.PROCESS0
T
nbriggs commented 1 month ago

Try (RETFROM 'OLDFAULT1 NIL)

pamoroso commented 1 month ago

Executing (RETFROM 'OLDFAULT1 NIL) from the debugger yields the error:

In OLDFAULT1:
DON'T is an undefined function.

The error context:

2/8_: BT
OLDFAULT1
NEWFAULT1
FAULTAPPLY
DON'T
DOUSERFNS2
SI::*UNWIND-PROTECT*
REDISPLAYW
MOVEW
ACE
FAULTEVAL
EVAL
EVAL-INPUT
DO-EVENT
XCL::EXECA0001A0002
XCL::EXECA0001
EXEC
T

2/9_: ?=
  FAULTX = DON'T
  FAULTARGS = ({WINDOW}#65,150000 (0 99 --) NIL --)
  FAULTAPPLYFLG = T
2/10_: PB FAULTX
bindings for FAULTX: 
 @ OLDFAULT1 : DON'T
 @ TOP : NOBIND
2/11_: PB FAULTARGS
bindings for FAULTARGS: 
 @ OLDFAULT1 : ({WINDOW}#65,150000 (0 99 --) NIL --)
 @ FAULTAPPLY : ({WINDOW}#65,150000 (0 99 --) NIL --)
 @ TOP : NOBIND
2/12_: PB FAULTAPPLYFLG
bindings for FAULTAPPLYFLG: 
 @ OLDFAULT1 : T
 @ TOP : NOBIND
2/13_: BTV
   FAULTX DON'T
   FAULTARGS ({WINDOW}#65,150000 (0 99 461 26) NIL NIL)
   FAULTAPPLYFLG T
OLDFAULT1
NEWFAULT1
   FAULTFN DON'T
   FAULTARGS ({WINDOW}#65,150000 (0 99 461 26) NIL NIL)
   DEF NIL
   %%LEXICAL-ENVIRONMENT%% NIL
   TRAN NIL
   TRANFN NIL
FAULTAPPLY
DON'T
   FNLST DON'T
   WINDOW {WINDOW}#65,150000
   ARG1 (0 99 461 26)
   ARG2 NIL
   ARG3 NIL
DOUSERFNS2
   SI::*CLEANUP-FORMS* SI::RESETUNWIND
SI::*UNWIND-PROTECT*
   WINDOW {WINDOW}#65,150000
   REGION (0 99 461 26)
   ALWAYSFLG NIL
   DSP #<Output Display Stream/71,106250>
   REPAINTFN DON'T
   CLIPREG (0 0 461 125)
   LISPXHIST ((&) (5 "2" . "_ ") "<not yet evaluated>" 
NIL SIDE (1 &))
   SI::*RESETFORMS* ((& NIL) (& NIL) (& NIL))
   RESETSTATE NIL
REDISPLAYW
   WINDOW {WINDOW}#65,150000
   POSorX NIL
   Y NIL
   OLDREGION (500 500 463 139)
   USERMOVEFN NIL
   OPEN? T
   OLDSCREEN {SCREEN}#65,151740
   POS (305 . 322)
   NEWREGION (305 322 463 139)
   OLDLEFT 500
   OLDBOTTOM 500
   OLDWIDTH 463
   OLDHEIGHT 139
   OLDCLIPREGION (0 0 299 99)
   LFT NIL
   BTM NIL
   REG NIL
   FN NIL
   NEWCLIPPINGREGION (0 0 461 125)
   NCL 0
   OCL 0
   NCB 0
   OCB 0
   OCR 299
   NCR 461
   OCW 299
   NCW 461
   OCH 99
   NCH 125
   OCT 99
   NCT 125
MOVEW
   SEQUENCE NIL
   WINDOW NIL
   POSITION NIL
   APPLICATION NIL
   FONT {FONTDESCRIPTOR}#74,73464
   TEMP.REGION NIL
ACE
   *FORM* (ACE)
   *ARGVAL* NIL
   *TAIL* NIL
   *FN* ACE
\EVALFORM
FAULTEVAL
   *FORM* (UNDOABLY (ACE))
\EVALFORM
   \INTERNAL NIL
EVAL
EVAL-INPUT
   RETRYFLAG NIL
   HELPCLOCK 626459364
DO-EVENT
   SI::*DUMMY-FOR-CATCH* T
   SI::*CATCH-RETURN-FROM* (&)
   LISPXHIST ((&) (5 "2" . "_ ") "<not yet evaluated>" 
NIL SIDE (1 &))
   HELPCLOCK 0
XCL::EXECA0001A0002
   *CURRENT-EVENT* ((&) (5 "2" . "_ ") 
"<not yet evaluated>" NIL SIDE (1 &))
   SI::NLSETQ-VALUE NIL
   *PROCEED-CASES* (&)
   SI::*NLSETQFLAG* NIL
XCL::EXECA0001
\PROGV
   XCL::TOP-LEVEL-P T
   XCL::WINDOW {WINDOW}#65,150150
   XCL::TITLE-SUPPLIED NIL
   XCL::TITLE NIL
   *THIS-EXEC-COMMANDS* (#<Hash-Table @ 70,37614>)
   XCL::ENVIRONMENT NIL
   XCL::PROMPT NIL
   XCL::FN EVAL-INPUT
   XCL::PROFILE "INTERLISP"
   *EXEC-ID* "2"
   XCL::PROFILE-CACHE NIL
EXEC
   *FORM* (EXEC :TOP-LEVEL-P T :PROFILE (QUOTE 
"INTERLISP") :ID (QUOTE NIL))
   *ARGVAL* NIL
   *TAIL* NIL
   *FN* EXEC
\EVALFORM
   *TAIL* ((EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
PROGN
   *FORM* (PROGN (TTYDISPLAYSTREAM &) (PROCESSPROP & & &
) (EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
\EVALFORM
   %#FORM# (PROGN (TTYDISPLAYSTREAM &) (PROCESSPROP & & &) (EXEC :TOP-LEVEL-P T :PROFILE & :ID &))
   *CURRENT-PROCESS* #<Process EXEC#2/74,107000>
   HELPFLAG BREAK!
   \CURRENTDISPLAYLINE 0
   \#DISPLAYLINES 31
   \LINEBUF.OFD #<IO Linebuffer Stream/74,23070>
   *READTABLE* #<ReadTable INTERLISP/74,146726>
   \PRIMTERMTABLE {TERMTABLEP}#74,141740
   \PRIMTERMSA {CHARTABLE}#74,142000
   TtyDisplayStream #<Output Display Stream/74,150070>
   SI::*RESETFORMS* NIL
   \INTERRUPTABLE T
   \TTYWINDOW NIL
   READBUF NIL
   \TERM.OFD #<Output Display Stream/71,106340>
   *STANDARD-OUTPUT* #<Output Display Stream/71,106340>
   *STANDARD-INPUT* #<IO Linebuffer Stream/74,23070>
\MAKE.PROCESS0
T
nbriggs commented 1 month ago

OK - since the code doesn't expect REDISPLAYW to be called (since there's a DON'T on it) - back up the stack and (RETFROM 'REDISPLAYW NIL) -- and if it calls REDISPLAYW again, have it return NIL again. I don't know if we have the Medley 1.x/2.x sources easily available, but looking at what the code in MOVEW is doing should enable you to identify why it called REDISPLAYW when it shouldn't have. I don't think the DON'T option was new.

pamoroso commented 1 month ago

Executing (RETFROM 'REDISPLAYW NIL) at the debugger yields the error:

In OLDFAULT1:
DON'T is an undefined function.

The backtrace confirms REDISPLAYW is being called again:

2/7_: (RETFROM 'REDISPLAYW NIL)
In OLDFAULT1:
DON'T is an undefined function.

2/8_: BT
OLDFAULT1
NEWFAULT1
FAULTAPPLY
DON'T
DOUSERFNS2
SI::*UNWIND-PROTECT*
REDISPLAYW
MOVEW
ACE
FAULTEVAL
EVAL
EVAL-INPUT
DO-EVENT
XCL::EXECA0001A0002
XCL::EXECA0001
EXEC
T

Executing (RETFROM 'REDISPLAYW NIL) again exits the debugger with no errors or warnings and displays this still partially corrupted ACE 2.1 window:

2/9_: (RETFROM 'REDISPLAYW NIL)
Animation Directory?  {DSK}NIL

ace-main1

Pressing ENTER finally displays the control window and its menu:

ace-main2

So ACE on Medley for DOS isn't affected by the issue reported here.

PRINT-LISP-INFORMATION summarizes the details of the sysout I'm running on DOS:

2/12_ (PRINT-LISP-INFORMATION)
Venue Medley version Medley 2.0 sysout of 24-Sep-92 15:41:04 on 386, 
Emulator created:  9-Feb-93, memory size: 0, machine NIL NIL based on 
Envos Medley version 
Medley 2.0 sysout of 24-Sep-92 15:41:04, Make-init dates: 24-Sep-92 13:53:02, 24-Sep-92 14:12:30
Patch files: 
NIL