taviso / 123elf Goto Github PK

Gentoo GNU/Linux (multilib), glibc-2.34

$ gdb -ex=r --args 123 -f unix -w $PWD/root/lotus/123.v10/smpfiles/income.wk3 
GNU gdb (Gentoo 11.2 vanilla) 11.2
Copyright (C) 2022 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-pc-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<https://bugs.gentoo.org/>.
Find the GDB manual and other documentation resources online at:
    <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from 123...
Starting program: /home/random/git/123elf/123 -f unix -w /home/random/git/123elf/root/lotus/123.v10/smpfiles/income.wk3
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".

Program received signal SIGSEGV, Segmentation fault.
0xf7e6f3c2 in __fstatat64_time64 () from /lib/libc.so.6
(gdb) backtrace
#0  0xf7e6f3c2 in __fstatat64_time64 () from /lib/libc.so.6
#1  0xf7e6f11d in fstatat () from /lib/libc.so.6
#2  0xf7e6ee40 in stat () from /lib/libc.so.6
#3  0x0812eab2 in __unix_stat (pathname=0x8142fa6 "/home/random", statbuf=0xffffc074) at wrappers.c:126
#4  0x0811d12a in path_exists ()
#5  0x081107ee in config_init ()
#6  0x08111fe7 in main_init ()
#7  0x08111f8e in __unix_main ()
#8  0x0812e839 in main (argc=5, argv=0xffffc664, envp=0xffffc67c) at main.c:19
(gdb) 

Also, I had to add -ltinfo to get it to link against ncurses 6.3:

diff --git a/Makefile b/Makefile
index 6ccea31..d3b93d0 100644
--- a/Makefile
+++ b/Makefile
@@ -4,7 +4,7 @@ OBJCOPY_FLAGS  = --wildcard --localize-symbols=localize.lst --globalize-symbols=
 CFLAGS  = -lc -m32 -ggdb3 -O0 -fno-stack-protector
 ASFLAGS = --32
 LDFLAGS = $(CFLAGS) -B. -Wl,-b,coff-i386 -no-pie
-LDLIBS = -lncurses
+LDLIBS = -lncurses -ltinfo
 PATH := .:$(PATH)
 
 define BFD_TARGET_ERROR

Hi

I've tried to build on

# lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 20.04.4 LTS
Release:        20.04
Codename:       focal

First build binutils with --prefix=/home/user/opt --enable-targets=all
And then

$ make
cc -m32 -ggdb3 -O0 -Wno-multichar  -m32 -ggdb3 -O0 -Wno-multichar -B. -Wl,-b,coff-i386  coffsyrup.c  -lncurses -o coffsyrup
./ld: cannot find -lncurses: No such file or directory
collect2: error: ld returned 1 exit status
make: *** [<builtin>: coffsyrup] Error 1

Installing every ncurses* pkg with 32-bit versions didn't help

# dpkg -l | grep curses | awk '{print $2}'
lib32ncurses6
libncurses-dev:amd64
libncurses5-dev:amd64
libncurses6:amd64
libncursesw5-dev:amd64
libncursesw6:amd64
mtr-tiny
ncurses-base
ncurses-bin
ncurses-term
pinentry-curses

Changing Makefile also didn't help
LDLIBS = -lncurses -L/usr/lib/x86_64-linux-gnu

$ make
cc -m32 -ggdb3 -O0 -Wno-multichar  -m32 -ggdb3 -O0 -Wno-multichar -B. -Wl,-b,coff-i386  coffsyrup.c  -lncurses -L/usr/lib/x86_64-linux-gnu -o coffsyrup
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libncurses.so.6 when searching for libncurses.so.6
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libncurses.so.6 when searching for libncurses.so.6
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libtinfo.so when searching for -ltinfo
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libtinfo.a when searching for -ltinfo
./ld: cannot find -ltinfo: No such file or directory
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libtinfo.so when searching for -ltinfo
./ld: skipping incompatible /usr/lib/x86_64-linux-gnu/libgcc_s.so.1 when searching for libgcc_s.so.1
collect2: error: ld returned 1 exit status
make: *** [<builtin>: coffsyrup] Error 1

Even with
LDLIBS = -lncurses -L/usr/lib32

$ make
cc -m32 -ggdb3 -O0 -Wno-multichar  -m32 -ggdb3 -O0 -Wno-multichar -B. -Wl,-b,coff-i386  coffsyrup.c  -lncurses -L/usr/lib32 -o coffsyrup
./ld: cannot find -lncurses: No such file or directory
collect2: error: ld returned 1 exit status
make: *** [<builtin>: coffsyrup] Error 1

I think we should make a .deb/.rpm, but how?

I think the main question is if it is okay to include a 123 binary, or should we include ld and link it together in a script?

I think /File List Worksheet doesn't seem to work?

It's hard to press some keys in some terminals, I've noticed F10 doesn't work in gnome-terminal.

I really need to figure out the keymap file format and make a utility that can edit it.

I think this is a slightly over-engineered file format that has a serialized n-ary tree of key sequences.

For example, PgUp might generate \e[6~ and → might generate \e[C

It could be stored like this:

                \e
                /\
               [  Escape
              /\
             6  C
            /\   \
           ~  ?  Right
           |
          PgUp

This way you navigate the tree to figure out what the user just pressed.

If you do /File Admin Table Worksheet it's supposed to add a list of files to your worksheet. It seems to work, but what is that third column supposed to be.... the timestamp? Is that a bug?

Saving is unfortunately not possible for me /File Save Enter name of file to save: /home/foo/file0001.wk3.

I get the following error message: Unable to reserve file: it is already reserved -- Press HELP

Unable to reserve file: it is already reserved

 You selected /File Admin Reservation Get when someone else was using
 the file with the reservation.

 You can get the reservation if no one else has it and if the file
 has not changed on disk since you read it into memory. Try again later
 to get the reservation.

This will create an empty file /home/foo/file0001.wk3.

To be on the safe side, I tried the following: /File Admin Reservation Release and get this error message The worksheet file has no reservation -- Press HELP.

The worksheet file has no reservation

 The following conditions could have caused this error:

 * You tried to save a file and you did not have the reservation. You must
   have the reservation to save a file. Use /File Admin Reservation Get
   to try to get the reservation.

 * You tried to release the reservation with /File Admin Reservation
   Release when you did not have the reservation. You are already
   using a read-only file.

 * You tried to save a worksheet file to a directory or disk file to which
   you do not have write-privileges at the UNIX level. Change the
   permissions with the UNIX  chmod  command and repeat the /File Save.

System: Ubuntu 22.04 64b
123elf version: 199b81f
Terminal: Konsole (same issue with xterm)

I made a successful binary using the provided scripts and makefile, but executing it just returns a "123: Cannot initialize system file configuration" error.

Trying to run it with gdb doesn't return any form of crash or error message, it terminates normally.

Maybe I skipped some configuration step, but the binary seems fine.

123 will start fine in a terminal window larger than 80x24. I normally run a 100x50 window and that is no problem at all. 100x80 also works well.

But when the window is larger, just slightly outside those dimensions, 123 fails on startup with the message "123: Memory full".

I'm not sure this can be changed, as we don't have the source code. But a note in the documentation would be helpful for those, like me, who run ridiculously large screens with ridiculously small fonts.

If the limit can be extended that would of course be even nicer. 123 runs perfectly fine in these larger terminal sizes otherwise.

I really can't understand from the instructions what to do.

I am running a Debian 11 system. The output of objdump --info | grep coff-i386 is blank. From this I assume I do not have a copy of binutils which supports coff-i386.

So what should I do next?

I won't go into explaining what I tried to do because this will just confuse matters.

The original command line parser of Lotus 1-2-3 for UNIX supported deprecated and useless features, and there are some outdated options as following:

• -f dos : select a file-name mode to DOS
• -c : select lower/upper-case dos file-name mode
• -n: run without using a video display

Meanwhile, we might extend the command line parser to support new options such as --no-banner / --show-banner (#28), --ignore-case (file-name mode ignoring the case of the characters), and --enable-autoexec / --disable-autoexec (#27).

These files cause a "segmentation fault":
wks-files.zip

This is the result:

This thing:

                           Lotus (R)   1-2-3 (R)   for {OS}
                                Version {PVERSION}

This is a video describing the issue:

https://photos.app.goo.gl/xrRVaprVntwpuuoHA

I configured it in VT100 emulation

Lots of keys don't work properly (e.g. F1).

Figure out why.

Sometimes Lotus says "Error reading file" on startup. The error can be dismissed by pressing Enter, and it seems to work after that.

What file is it talking about?

Lotus 1-2-3 predates UTF-8, and uses LMBCS internally, which is sort of a precursor to unicode.

I see no reason we couldn't add a UTF-8 option for file/display charset, for better i18n support. It supports character set translation, we just have to teach it how and figure out the CBD (character bundle) format. I already know the BDLREC format, from my lotusdrv project - it's basically a TLV (tag, length, value) encoding system.

Running make gave me multiple definition of 'ustat'

./ld: 123.o:ulimit.s:(.text+0xdfa0c): multiple definition of `ustat'

Adding ustat to undefine.lst allowed make to complete.

Ubuntu 18.04

Hello!

Trying to compile on Ubuntu 20.04 using binutils-2.34 downloaded and compiled with i386-coff support. Getting the following:

objcopy -I coff-i386 -O coff-i386 --wildcard --localize-symbols=localize.lst --globalize-symbols=globalize.lst --redefine-syms=redefine.lst --weaken-symbols=weaken.lst orig/123.o 123.o
free(): double free detected in tcache 2
make: *** [Makefile:44: 123.o] Aborted (core dumped)
make: *** Deleting file '123.o'

I'm thinking this might be the result of the version of binutils I'm using? It's the same version as Ubuntu 20.04 gives me. Any ideas?

Thanks!

You can create a graph, but can't display it in the terminal yet -- port the driver from lotusdrv!

(more of a question for you related to your article on lotus123)
Not sure if this is the kind of code you are looking for a COFF object loader, I once wrote one based off a similar one we had working in good old sunos3.... All those codes are in the NEMO package in:
https://github.com/teuben/nemo/tree/master/src/kernel/loadobj

Hi, thank you for your amazing work!
I am trying to build on WSL / Debian 10 but got the following error:

cc forceplt.o -lc -m32 -ggdb3 -O0 -fno-stack-protector -lc -m32 -ggdb3 -O0 -fno-stack-protector -B. -Wl,-b,coff-i386 -no-pie 123.o dl_init.o main.o wrappers.o patch.o filemap.o graphics.o draw.o -Wl,--whole-archive,ttydraw/ttydraw.a,--no-whole-archive -o 123 -lncurses -ltinfo
./ld: forceplt.o: in function __require_ref': (.text+0x470): warning: the use of tmpnam' is dangerous, better use mkstemp' ./ld: (.text+0x466): warning: the use of tempnam' is dangerous, better use mkstemp' ./ld: 123.o:atoi.c:(.text+0xd9f9c): multiple definition of getmsg'
./ld: 123.o:putchar.c:(.text+0xdc0a4): multiple definition of putmsg' ./ld: 123.o:setjmp.s:(.text+0xdd990): multiple definition of stime'
collect2: error: ld returned 1 exit status
make: *** [Makefile:44: 123] Error 1

Thanks!

123 (built at 2b15de4) runs one core at 100%. stracing it shows it’s polling stdin for characters:

times({tms_utime=0, tms_stime=0, tms_cutime=0, tms_cstime=0}) = 1721980973
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, SNDCTL_TMR_START or TCSETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
ioctl(0, TCGETS, {c_iflags=0x4003, c_oflags=0x4, c_cflags=0xbf, c_lflags=0xa31, c_line=0, c_cc[VMIN]=0, c_cc[VTIME]=0, c_cc="\x03\x1c\x7f\x15\x04\x00\x00\x00\x11\x13\x1a\x00\x12\x0f\x17\x16\x00\x00\x00"}) = 0
read(0, "", 1)                          = 0

continually shows up.

I’m using Debian 11 with XTerm 366.

/System is supposed to drop you to a command shell, useful for macros.

It almost works... sometimes, but not quite and the terminal messes up.

https://github.com/andmarti1424/sc-im

It would be nice to a have script that opens a bunch of worksheets and checks everything is working so that we're not scared we're breaking something with big changes.

Given all the specific dependencies (and the main repo size, for instance), may be providing an AppImage would a nice shortcut for those who want to try it out?

I'm not saying in any way that the instructions are limited or something, not at all -- I've built the package without any problems on Ubuntu 21.10. It's more like just a thought may be :)

In any case, thanks for bringing the memories back, really great work!

Ports to BSD variants should not be much more difficult than Linux.

Currently build fails on NetBSD:

cc -m32 -ggdb3 -O0 -fno-stack-protector  -D_FILE_OFFSET_BITS=64 -D_TIME_BITS=64 -D_GNU_SOURCE -I ttydraw -c wrappers.c
wrappers.c: In function ‘__unix_ioctl’:
wrappers.c:62:14: error: ‘TCSETSW’ undeclared (first use in this function); did you mean ‘TC_TSS’?
         case TCSETSW:
              ^~~~~~~
              TC_TSS
wrappers.c:62:14: note: each undeclared identifier is reported only once for each function it appears in
wrappers.c:67:14: error: ‘TCSETS’ undeclared (first use in this function); did you mean ‘TCSETSW’?
         case TCSETS:
              ^~~~~~
              TCSETSW
wrappers.c:83:14: error: ‘TCGETS’ undeclared (first use in this function); did you mean ‘TCSETS’?
         case TCGETS:
              ^~~~~~
              TCSETS
wrappers.c: In function ‘__unix_readdir’:
wrappers.c:407:26: error: ‘struct dirent’ has no member named ‘d_off’
         uent.d_off = lent->d_off;
                          ^~
wrappers.c: At top level:
wrappers.c:419:1: error: unknown type name ‘sighandler_t’; did you mean ‘sa_handler’?
 sighandler_t __unix_signal(int signum, sighandler_t handler)
 ^~~~~~~~~~~~
 sa_handler
wrappers.c:419:40: error: unknown type name ‘sighandler_t’; did you mean ‘sa_handler’?
 sighandler_t __unix_signal(int signum, sighandler_t handler)
                                        ^~~~~~~~~~~~
                                        sa_handler

If you're trying to back out of several menus by pressing the escape key a few times, 1-2-3 seems very slow. It's actually just waiting a second to see if the Esc is the first character in a sequence.

I think this timeout is too high for modern computers, this was probably written when you might using 123 over a lossy 2400 baud modem or something. In an xterm, we can make the timeout tiny.

Using commit 199b81f.

I ran git pull, then make clean and then make, then copied the l123set file to my home directory before trying to run the program again.

Terminal size 80x24: Works fine
Terminal size 300x70 (zsh says it's an exit code 139):

                   -`                    jeff@archpad 
                  .o+`                   ------------ 
                 `ooo/                   OS: Arch Linux x86_64 
                `+oooo:                  Host: 20BWS0PJ00 ThinkPad T450s 
               `+oooooo:                 Kernel: 5.17.9-arch1-1 
               -+oooooo+:                Uptime: 16 mins 
             `/:-:++oooo+:               Packages: 626 (pacman) 
            `/++++/+++++++:              Shell: zsh 5.9 
           `/++++++++++++++:             Resolution: 1920x1080 
          `/+++ooooooooooooo/`           WM: JWM 
         ./ooosssso++osssssso+`          Theme: Raleigh [GTK2/3] 
        .oossssso-````/ossssss+`         Icons: Papirus [GTK2/3] 
       -osssssso.      :ssssssso.        Terminal: xterm 
      :osssssss/        osssso+++.       CPU: Intel i7-5600U (4) @ 3.200GHz 
     /ossssssss/        +ssssooo/-       GPU: Intel HD Graphics 5500 
   `/ossssso+/:-        -:/+osssso+-     Memory: 1173MiB / 11843MiB 
  `+sso+:-`                 `.-/+oso:
 `++:.                           `-/+/                           
 .`                                 `/                           


jeff@archpad ~/builds/123elf (git)-[main] % ./123
[1]    5338 segmentation fault (core dumped)  ./123
                                                   %                          
139 jeff@archpad ~/builds/123elf (git)-[main] %                            :( 

Output from neofetch is included in case it helps in some way.

Hi! I built binutils to support coff-i386, got all the dependencies lined up and compilation to succeed. I've gotten as far as:

[=D][josefcub@sinir 123elf]$ ./123
Segmentation fault (core dumped)

I noticed that some options do work:

[=P][josefcub@sinir 123elf]$ ./123 -h
Usage: 123 [optional argument]
Optional arguments are:
        -f unix or -f dos       to select a file-name mode
        -c lower or -c upper    to select lower-case dos file-name mode
                                  or upper-case dos file-name mode
        -k keyboard-type        to select a particular keyboard type
        -n                      to run without using a video display
        -p number               to select the number of sheets displayed
                                  in perspective mode
        -w filename             to select a file to override the auto-file

(-n works as expected; the rest lead to a segfault). When I ran strace, I ended up with this:

openat(AT_FDCWD, "/proc/self/cwd/root/lotus/123.v10/cbd/l_ascii.bun", O_RDONLY) = 4
statx(4, "", AT_STATX_SYNC_AS_STAT|AT_NO_AUTOMOUNT|AT_EMPTY_PATH, STATX_BASIC_STATS, {stx_mask=STATX_BASIC_STATS|STATX_MNT_ID, stx_attributes=0, stx_mode=S_IFREG|0644, stx_size=2910, ...}) = 0
read(4, "\\\v]\0\f\4-\0\4\0\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377"..., 2910) = 2910
close(4)                                = 0
ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
--- SIGSEGV {si_signo=SIGSEGV, si_code=SI_KERNEL, si_addr=NULL} ---
rt_sigaction(SIGSEGV, {sa_handler=SIG_DFL, sa_mask=[SEGV], sa_flags=SA_RESTART}, {sa_handler=0x811a438, sa_mask=[SEGV], sa_flags=SA_RESTART}, 8) = 0
getpid()                                = 31626
kill(31626, SIGSEGV)                    = 0
sigreturn({mask=[]})                    = 24
--- SIGSEGV {si_signo=SIGSEGV, si_code=SI_USER, si_pid=31626, si_uid=1000} ---
+++ killed by SIGSEGV (core dumped) +++

I'm not much of a developer beyond being able to get this far, but I can provide other information (installed package versions, etc) if you need them. I did change my TERM variable around, and even ran it from the Linux console, to see if TCGETS was hitting it with something that Konsole was providing, but no dice.

I realize you're on Fedora, and I'm on something a little different, so I'm just reporting this in the spirit of 'this might break soon for more than just me'.

I've included the whole strace, in case it provides something I'm not seeing:
strace.txt
(The strace output differs here, probably due to 1>strace.txt 2>&1 )

Thank you for reading, and thanks for putting this really neat package together!

In extract.sh, the find ... gunzip step fails with

invalid compressed data--code out of range

for many files using gunzip 1.6 on Ubuntu 18.04. I tried a few other ways to extract the .z files on my Linux machine, but the one that worked was using gunzip 272.250.1 on MacOS.

I would love to give this a test drive out of curiosity.
But it doesn't look easy for me to build this, or is it?
It would be great to have a Dockerfile or docker image so that running this would be super easy.
Maybe someone is already working on this?

binutils.sh fails, if the path to the directory contains a blank.

Example:
Compiling binutils in this directory fails:
/home/user/Documents/My Documents/123elf

Compiling binutils in this directory works for me (on the same system, Debian Bullseye 11):
/home/user/Documents/123elf

• There seems to be some border on lower right part of pie charts, is that a bug or is it supposed to be a shadow?
• The hatching options don't work very well
• Footnotes work, but can be clobbered if the labels are too close
• Can I improve appearance with Unicode box drawing characters?
• The font options don't work, but we could make them work with figlet or unicode or something.

Using "/" menu, and pressing ESC, it takes about one second before exiting the menu and return to the edit mode on spreadsheet. It happens on Terminal Emulator in Cinnamon:

Is it normal?

P.S. it does not happen on a real DEC VT320...

I had to install makeinfo

sudo apt install texinfo

before the binutils.sh would complete. There was a failure during make.

The script completed successfully after makeinfo was installed.

Ubuntu 18.04

coffsyrup cannot currently undefine local symbols, but there's no reason why it can't. The main problem is there are no relocs, so existing callers will still call the old function. There is an easy solution:

• undefine the symbol
• replace the first instruction with jmp [undefined symbol]

Now we provide a replacement, and the linker will just fix it up for us.

The print menus are there. I have a default printer. Its not able to spool a print job to it (or anywhere else) as fas I can tell. I could try setting up some text style old world 'lpr' printers perhaps. Print to File segmentation faults.

Tried compiling with the official instructions on Arch Linux and get this error when I run ./123:

123: setup file </home/ironveil/.l123set> damaged, please run setup123

Not sure what went wrong, even after a fresh recompile it still persists. I can add any more information if needed

MacOS cpio doesn't have the -D directory option so I used a relative reference in the command below.

When I did

cd root
for i in 123UNIX{2..5}.IMG; do
    cpio -id < ../${i}
done

The lotus/123.v10/sysV386/lib/123.o.z_1 file (and others) extracted successfully, but 123.o.z_2 did not. I don't see any reason that this happens except that on Linux I get

lotus/123.v10/hlp/USA-English/123_sysV.hlp.z
cpio: warning: skipped 4798 bytes of junk
lotus/123.v10/sysV386/dvr/l13pepdt.dvr.z
lotus/123.v10/sysV386/lib/123.o.z_2
2510 blocks
lotus/123.v10/sysV386/bin/inst_dl.z

when I use -v. On MacOS, I get skipped 4798 bytes... and the _2 file isn't listed in the file names.

It seems surprising that Autoexec is enabled by default. I think this should be off by default, and users can enable it via /Worksheet Global Default Autoexec.

I get an "WAIT" then an "ERROR" message on top right of the screen when I try to save a worksheet:

WAIT:

ERROR:

When the spreadsheet is on the screen, press sequentially these keys:

• ESC
• [
• a

The result will be an abnormal program termination as per this screenshot:

I think modern users will be surprised that you have to enable undo (/Worksheet Global Default Other Undo Enable Update).

Currently program doesn't start if not invoked from the code repo directory:

$ pwd
/
$ /tmp/123elf/123 -f unix -w "/tmp/123/root/lotus/123.v10/smpfiles/income.wk3"
123: can't access directory </proc/self/cwd/root/lotus/123.v10>, errno = 13
123: impossible d'acceder au repertoire </proc/self/cwd/root/lotus/123.v10>, erreur = 13

We should have a make install script that will, roughly:

• copy the binary into /usr/local/bin/123 (where user can change the /usr/local prefix)
• copy the man page into /usr/local/man/man1/123.1
• copy other needed files from root/lotus/ into e.g. /usr/local/share/lotus/, and the program should be able to find them there

If you press F2 to edit a very long formula, then move around with the cursor keys, backspace and del don't always seem to work correctly. I think this is a termcap issue.

In 123R4 you can do @DATE(2022, 5, 22). 123 for UNIX only accepts two digit years, it seems to display okay (use /Range Format Date), but I'm not sure if dates spanning Y2K will sort properly?

This can probably just be patched so that @DATE() works.

I really need to make sure all the other functions are Y2K compliant 😆

I am using Ubuntu 20.04 and managed to compile it, however it segfaults right after the initial popup showing license information.

Compilation output:

$ make
cc forceplt.o -lc -m32 -ggdb3 -O0 -fno-stack-protector -lc -m32 -ggdb3 -O0 -fno-stack-protector -B. -Wl,-b,coff-i386 -no-pie 123.o dl_init.o main.o wrappers.o patch.o -o 123 -lncurses
./ld: forceplt.o: in function `__require_ref':
(.text+0x470): warning: the use of `tmpnam' is dangerous, better use `mkstemp'
./ld: (.text+0x466): warning: the use of `tempnam' is dangerous, better use `mkstemp'
./ld: (.text+0x47a): warning: gtty is not implemented and will always fail
./ld: (.text+0x47f): warning: stty is not implemented and will always fail
123: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux.so.2, BuildID[sha1]=284e6f6d00cb478dfe08fbabf732df59e61a4507, for GNU/Linux 3.2.0, with debug_info, not stripped
   text	   data	    bss	    dec	    hex	filename
 943897	 136840	  41328	1122065	 111f11	123
Copy l123set.cf to ~/.l123set and 1-2-3 should be ready!

GDB output:

Reading symbols from ./123...
(gdb) run
Starting program: ~/123elf/123 

Program received signal SIGSEGV, Segmentation fault.
0x08127031 in getuid@GLIBC_2.0 ()
(gdb) bt
#0  0x08127031 in getuid@GLIBC_2.0 ()
#1  0x08112093 in __unix_main ()
#2  0x0812e987 in main (argc=1, argv=0xffffd5e4, envp=0xffffd5ec) at main.c:19

When using gdb it jumps straight into the SIGSEGV.

I have tried to uncomment that syscall in redefine.lst but then it crashes on another call:

Program received signal SIGSEGV, Segmentation fault.
0x08127031 in gets@GLIBC_2.0 ()
(gdb) bt
#0  0x08127031 in gets@GLIBC_2.0 ()
#1  0x0812e987 in main (argc=1, argv=0xffffd5e4, envp=0xffffd5ec) at main.c:19

Any ideas?

Add an option to disable the welcome banner on startup, it does sleep(5) in banner_on() which can be annoying.

banner_on() is very simple and can just be rewritten, perhaps it should do getenv("LOTUS_NO_BANNER") or something.