Kent,

As a new user of KEGS on a brand new Intel i7 Mac, all I can say is WOW!

It absolutely flies. Tell me, while KEGS shows 500-600 MHz under GS/OS,
when I run the ProDOS 8 AppleWorks 5.1 it shows 4500 MHz. Is that even
possible? If so, Holy ZipChip, Batman!

Today was my first day of using KEGS on the Mac (though I still use the
KEGS-based GSPort on a Windows 10 machine) so I'm still getting a feel
for things.

Direct printing has always been big deal to me, so I'm going to set up
an ncat client-to-client relay that will run while KEGS is running to
funnel your 6501 socket output on the printer port to a networked
LaserJet printer's port 9100 (raw) and make sure that will work for me
in the Mac environment. (It does work with GSPort on Windows for both
PCL and PostScript jobs, so I imagine it will work here too).

I suppose a more direct approach would be for me to ask you to implement
an outgoing socket client (not the modem on telnet port) for the printer
port where the IP address and port # of the network printer server could
be specified, but that's down the road. The printer server disconnects
after a short idle period, so the outgoing client socket would need
automatically to reconnect periodically, which might complicate things,
I imagine.

One other thing -- regarding your Dynapro file typer -- it would be most
appreciated if, on the Mac, you implemented the legacy 4-character Mac
type to denote the Apple II file type and aux type of files in the
Dynapro folder. (e.g. pxyy / x=FT yy=AT - Programmer's Reference for
[GS/OS] System 6.0 (page 335-336)). I realize that not many MacOS
programs still use the legacy type info, but it's still supported in the
file system. And, I know Kelvin Sherlock's ShrinkFit-X and Andy
McFadden's nulib2 still read and write that info on MacOS.

In any case -- Really, thanks for this program. What a blast.

Hugh Hood

On Wednesday, June 14, 2023 at 9:10:31 PM UTC-7, Hugh Hood wrote:
> One other thing -- regarding your Dynapro file typer -- it would be most
> appreciated if, on the Mac, you implemented the legacy 4-character Mac
> type to denote the Apple II file type and aux type of files in the
> Dynapro folder. (e.g. pxyy / x=FT yy=AT - Programmer's Reference for
> [GS/OS] System 6.0 (page 335-336)). I realize that not many MacOS
> programs still use the legacy type info, but it's still supported in the
> file system. And, I know Kelvin Sherlock's ShrinkFit-X and Andy
> McFadden's nulib2 still read and write that info on MacOS.

Code for the file type conversion can be found here:

https://github.com/fadden/CiderPress2/blob/984958b4deaf827ab0ff1e7c6e9885f1b5ceb9c4/DiskArc/FileAttribs.cs#L210

The cp2 command-line tool will use the HFS+ type/creator fields (and resource forks) when adding and extracting files, roughly the same way NuLib2 does.

I also say is WOW!

As an old user of KEGS on a few-years-old Intel i7 Dell laptop running Linux
Mint with more than 280 processes and all but the KEGS process sleeping.
Press F6 enough times to set the KEGS speed set to Unlimited, and the KEGS
process is under 50% of the CPU usage:

$ ps -C kegs -o pcpu

%CPU
32.8

I rarely look at those numbers, but look ...

When at the Applesoft prompt:

dfcyc: ~1280
sim MHz: ~9600
Eff NHz: ~4200
Limit: Unlimit

When running this test program:

*300:AD 57 C0 EA AD 56 C0 4C 0 3 N300G

dfcyc: ~460
sim MHz: ~28000
Eff NHz: ~4100
Limit: Unlimit

What do those numbers mean?

~mmphosis

ps. I also use the socket/serial port feature a lot but not for printing.
https://mmphosis.netlify.app/kegs/alias-t.html

In article <mmphosis-1686866335@macgui.com>,
mmphosis <mmphosis@macgui.com> wrote:
>I also say is WOW!
>
>As an old user of KEGS on a few-years-old Intel i7 Dell laptop running Linux
>Mint with more than 280 processes and all but the KEGS process sleeping.
>Press F6 enough times to set the KEGS speed set to Unlimited, and the KEGS
>process is under 50% of the CPU usage:
>
>$ ps -C kegs -o pcpu
>
>%CPU
>32.8
>
>
>I rarely look at those numbers, but look ...
>
>When at the Applesoft prompt:
>
>dfcyc: ~1280
>sim MHz: ~9600
>Eff NHz: ~4200
>Limit: Unlimit
>
>When running this test program:
>
>*300:AD 57 C0 EA AD 56 C0 4C 0 3 N300G
>
>dfcyc: ~460
>sim MHz: ~28000
>Eff NHz: ~4100
>Limit: Unlimit
>
>What do those numbers mean?
>
>~mmphosis
>
>ps. I also use the socket/serial port feature a lot but not for printing.
>https://mmphosis.netlify.app/kegs/alias-t.html

KEGS always slows down to 1MHz for accessing softswitches, like the
keyboard at $C000, or $C057, etc. Outside of that, KEGS tries to run as
fast as possible in Unlimited mode. So if the host machine runs fast
enough that the softswitch operation being requested can be done more
than 1 million times per second, then the KEGS speed effectively becomes
infinite--it did the 1 million accesses to the softswitch in a second,
had time left over, so it is basically dividing by 0 to get the "sim
MHz" and "eff MHz".

KEGS on Windows also has a bug I haven't chased down yet where it doesn't run
as fast as it could for some reason, it's confused about time on Windows
and puts itself to sleep when it does not need to, unrelated to this
one-million-softswitch accesses per second limit. This seems to be about
a 30-40% slowdown, but it doesn't always happen. So Windows users are
less likely to get "infinite" MHz numbers, since KEGS has other issues there.

KEGS is fast enough that the keyboard reading at most prompts (reading
$C000 from BASIC) can be done more than 1 million times per second, so
the speed KEGS shows at most prompts is not meaningful. Note lots of
other instructions are being run, maybe 100 65816 instructions per read of
$C000, but since KEGS is fast enough, it can do those in the "spare" time.

I guess I should consider changing KEGS to NOT slow down for softswitches
when running at Unlimited speed.

KEGS's real speed can be seen doing something, such as:

10 A=SIN(A) + 1
20 GOTO 10
RUN

(If KEGS begins to run faster than 3GHz, even BASIC will get caught by
the read to $C000 checking for Ctrl-C between each statement, and then even
running BASIC the speed will be "infinite", but really around 3GHz. This
will not be a problem anytime soon, KEGS runs at 1GHz max now on Apple M1,
and around 750-850MHz on Intel/AMD chips)

When running at 1MHz and 2.8MHz, KEGS also slows down to 1MHz for writes to
shadowed video memory (hires, lores, text, whatever, even if it's not
being displayed currently). But at 8MHz and Unlimited, it does not,
basically to avoid this "infinite MHz" speed issue when accessing video
memory.

What the fields mean: dfcyc is counting simulated seconds (which should
match real seconds, ignoring times when you halted KEGS). "sim MHz"
is the time KEGS spent just running CPU instructions and how fast it was,
not counting overhead for drawing video or playing sounds. "eff MHz" takes
the overhead into account, and is a truer score. So "sim MHz" should be
higher than "eff MHz", a lot higher if there's a lot of graphics or sound
going on. Both are affected by the one-million-softswitch accesses per
second limit, and so can scale too high when code is reading softswitches
aggressively.

Kent

In article <olKdnfqmfPKwEhf5nZ2dnZfqnPWdnZ2d@earthlink.com>,
Hugh Hood <hughhood@earthlink.net> wrote:
>Kent,
>
>As a new user of KEGS on a brand new Intel i7 Mac, all I can say is WOW!
>
>It absolutely flies. Tell me, while KEGS shows 500-600 MHz under GS/OS,
>when I run the ProDOS 8 AppleWorks 5.1 it shows 4500 MHz. Is that even
>possible? If so, Holy ZipChip, Batman!

As I posted later in this thread, the 4500MHz is not real. The emulated
system is reading the keyboard too often with nothing else, and KEGS is
not able to calculate the speed properly.

>Today was my first day of using KEGS on the Mac (though I still use the
>KEGS-based GSPort on a Windows 10 machine) so I'm still getting a feel
>for things.
>
>Direct printing has always been big deal to me, so I'm going to set up
>an ncat client-to-client relay that will run while KEGS is running to
>funnel your 6501 socket output on the printer port to a networked
>LaserJet printer's port 9100 (raw) and make sure that will work for me
>in the Mac environment. (It does work with GSPort on Windows for both
>PCL and PostScript jobs, so I imagine it will work here too).
>
>I suppose a more direct approach would be for me to ask you to implement
>an outgoing socket client (not the modem on telnet port) for the printer
>port where the IP address and port # of the network printer server could
>be specified, but that's down the road. The printer server disconnects
>after a short idle period, so the outgoing client socket would need
>automatically to reconnect periodically, which might complicate things,
>I imagine.

You want an option where the slot 1 port is set to auto-connect to, say,
10.0.0.10 port 1000 whenever there is outgoing traffic, right? I'll look
into what this would take.

>One other thing -- regarding your Dynapro file typer -- it would be most
>appreciated if, on the Mac, you implemented the legacy 4-character Mac
>type to denote the Apple II file type and aux type of files in the
>Dynapro folder. (e.g. pxyy / x=FT yy=AT - Programmer's Reference for
>[GS/OS] System 6.0 (page 335-336)). I realize that not many MacOS
>programs still use the legacy type info, but it's still supported in the
>file system. And, I know Kelvin Sherlock's ShrinkFit-X and Andy
>McFadden's nulib2 still read and write that info on MacOS.

As I understand it, you want KEGS to support native Mac Creator types--which
are stored in a weird way on a Mac (either hidden, or in .MACOS_X directories
or xattr or whatever, this is a huge giant mess of complexity which I do not
understand). I am not planning on doing that. Dynapro uses ,tbin,a$2000
extensions, and supports Cadius's #062000 extension as well. I'd be happy
to support anythin else that is portable to Linux and Windows, but the
Mac "hidden" filetype stuff is not.

>In any case -- Really, thanks for this program. What a blast.
>
>
>
>
>Hugh Hood

Thank you for the nice words.

Kent

On 6/16/2023 12:24 PM, Kent Dickey wrote:
>
> You want an option where the slot 1 port is set to auto-connect to, say,
> 10.0.0.10 port 1000 whenever there is outgoing traffic, right? I'll look
> into what this would take.
>

Exactly.

For example, here is the test application -- most mid and upper end
networked laser printers support the PCL 5 and PostScript languages, and
their print servers will accept raw input over Port 9100. Most also will
auto-detect whether the language being sent for that job is PCL 5 or
PostScript.

So, from GS/OS applications one would select the Direct Connect
PostScript driver and be able to print directly from GS/OS within KEGS
to the networked printer on the host (Mac/Windows/etc) machine.

And, from ProDOS 8 applications, one would print from BASIC and like
text-output programs because PCL 5 accepts plain ASCII as its input even
without attaching any PCL 5 specific language commands to it.

Moreover, from programs like AppleWorks 5 that have a built-in PCL 5
driver, one could print directly from AppleWorks 5 under ProDOS 8 within
KEGS straight to the networked printer, and use all the PCL features
such as font size selection, bold/non-bold, italic/non-italic, etc.

I currently do this very thing on my Windows 10 machine from the GSPort
variant of your KEGS, but I have to start an ncat client-to-client relay
bridging Port 6501 (from KEGS) and Port 9100 (on the printer) and keep
it looping in the background, because the Port 9100 server on the
printer will disconnect after a period of inactivity. It would be nice
not to have to do that.

So, in a nutshell, that is my reason for requesting this feature addition.

One more thing -- I was out of town this past weekend and was unable to
do any more testing with my newly-installed KEGS on the Mac, but using a
telnet window to connect to KEGS' port 6501, and then typing PR#1 in
BASIC on the emulated Apple IIGS would indeed transfer any and all
keystrokes typed on the emulated Apple IIGS to the telnet window, but it
would not, for example, transfer the contents of the Catalog command
typed from the same prompt.

May I ask you to check that out on your machine?

BTW, thanks for the quick response, and the damn-quick emulator.

Hugh Hood

Hi Hugh,

I am running version 1.23 on Linux. I can telnet in from the Terminal
program/Application on both Linux and Mac OS X. I don't have a "modern" Mac
to test the latest version of KEGS on, but it all works on KEGS 0.91 on Mac
OS X 10.5.8 as well.

From Linux locally:

telnet localhost 6501

From the PowerPC Mac remotely:

telnet 10.0.0.101 6501

On the KEGS side:

KEGS Configuration F4

↲ Serial Port Configuration

Serial Ports = Only use sockets 6501-6502
Serial Output = Mask off high bit
√ Modem on port 0 (slot 1) = Simple socket emulation mode

for both input and output, I type:

IN#1
PR#1

I boot DOS 3.3, and the CATALOG command outputs fine to the Terminal. Can
you print something from BASIC?

?"HELLO

Maybe newer macos has a security setting blocking the port? Or maybe telnet
needs something else set, it may be that it is simply echoing what you are
typing in telnet but isn't receiving and/or outputting the incoming text. I
used to set mode character from telnet, but it doesn't seem necessary. When
in telnet, press control+] to get to the telnet prompt:

telnet> mode character

~mmphosis
https://mmphosis.netlify.app/kegs/alias-t.html

On 6/20/23 1:30 PM, mmphosis wrote:

Thanks for the input. I'm still getting my feet wet with this.

>
> I am running version 1.23 on Linux. I can telnet in from the Terminal
>

This is KEGS version 1.27 on MacOS 10.14.6 (Mojave).

>
> I boot DOS 3.3, and the CATALOG command outputs fine to the Terminal. Can
> you print something from BASIC?
>
> ?"HELLO
>

All I can get from BASIC are the keystrokes typed while PR#1 is active.
I get no output from the CATALOG command nor the LIST command nor can I
print from BASIC (or AppleWorks for that matter).

My serial port settings are the same as yours.

OTOH, I can fire up GSPort with the same (as on my KEGS Mac) serial port
settings and Telnet command on a Windows 10 machine and everything comes
through with flying colors -- output of CATALOG, LIST, BASIC and even
AppleWorks.

>
> Maybe newer macos has a security setting blocking the port? Or maybe telnet
> needs something else set, it may be that it is simply echoing what you are
> typing in telnet but isn't receiving and/or outputting the incoming text.
>

That's a good thought. Thanks. I'll continue to experiment with this
until either Kent says something's wrong or I figure it out. I really
like this emulator.

Hugh Hood

I just checked my security settings in Mac OS X 10.5.8. They've moved things
and changed things around a bit in the newer macos, but there should be a
way:

System Preferences > Security > Firewall (tab) > Set access for specific
services and applications >
KEGSMAC Allow incoming connections

Turn this on if you are behind another firewall/router between your Mac and
the Internet, or better yet temporarily disconnect your Mac from the
Internet. You don't want someone telnetting in and typing INIT HELLO. I
would advise reverting this security setting when not using it.

I also downloaded version 1.27 and ran it on Linux and the telnet stuff
works. Are you using telnet from the Terminal Application, or are you using
a different Telnet Application altogether?

I really like being able to Command+C and Command+V between a
Text/Programming Editor and the Terminal connected to KEGS via telnet.
~mmphosis

You are correct — with KEGS version 1.27, the serial port stuff is not
working. I forgot to do "make clean" before building version 1.27. I think
this is why I was still using version 1.23.

There is a bit of output from telnet in the Terminal but it is garbled:

?S]CATALOG
VO*A 00 DE 04 TE TE*I00 Y. 0 UM 0 *A 27 T1 *A

This is the output from running kegs in the Terminal:

scc socket close being called from socket_open_out
In scc_socket_close, 0, 0, 00000180c9170000
sockfd ret: a
SCC port 0 is at unix port 6501
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
Set port[0].rdwrfd = b
Telnet reqwill and reqdo's initialized
In modem_do_ring, ringing at 000001bda7b40000
Telnet reqwill 1
Telnet reqdo 1
Telnet reqwill 3
Telnet reqdo 3
Send telnet reqs

and lots of ...

tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1
tx_dcycs < 1

On 6/20/23 3:03 PM, mmphosis wrote:
> You are correct — with KEGS version 1.27, the serial port stuff is not
> working. I forgot to do "make clean" before building version 1.27. I think
> this is why I was still using version 1.23.
>
> There is a bit of output from telnet in the Terminal but it is garbled:
>
> ?S]CATALOG
> VO*A 00 DE 04 TE TE*I00 Y. 0 UM 0 *A 27 T1 *A
>

Thank you, thank you, thank you. That output from the CATALOG command
was *exactly* what I was seeing.

FWIW, I downloaded the older (and slower) GSPort and serial output is
working as it should.

The solution is nigh, methinks.

Hugh Hood

Here's more of the output from kegs version 1.27 in the terminal:

Speaker at rest
GRIND GRIND GRIND
c7_fd == -1!
Failure reading boot disk in s7d1!
Pausing sound
Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!
Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!
scc socket close being called from socket_open_out
In scc_socket_close, 0, 0, 0000008773070000
sockfd ret: aWr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!
Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!

SCC port 0 is at unix port 6501
tx_dcycs 0.28 < 1 1 0
tx_dcycs 0.28 < 1 1 0
Set port[0].rdwrfd = b
Telnet reqwill and reqdo's initialized
In modem_do_ring, ringing at 000000b3d23a0000
Telnet reqwill 1
Telnet reqdo 1
Telnet reqwill 3
Telnet reqdo 3
Send telnet reqs
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18
tx_dcycs 0.28 < 1 19 18

I modified scc.c to get a little more information:

/* do just one char, then get out */
//printf("tx_dcycs < 1\n");
printf("tx_dcycs %g < 1 %d %d\n", scc_ptr->tx_dcycs, out_wrptr, out_rdptr);
return;

I am not too familiar with this socket code, but these lines don't appear in
kegs version 1.23:

Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!
Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!

Here's the output from kegs 1.23 that works:

Writing config.kegs file to config.kegs
GRIND GRIND GRIND
c7_fd == -1!
Failure reading boot disk in s7d1!
Partition error: No driver descriptor map found
Inserting disk dos33.dsk ((null) or -1) in slot 6, drive: 0, dyna_blocks:0
Writing config.kegs file to config.kegs
Toggling g_limit_speed to 1...1.024MHz!
Pausing sound
scc socket close being called from socket_open_out
In scc_socket_close, 0, 0, 000000a872730000
sockfd ret: 10
SCC port 0 is at unix port 6501
Set port[0].rdwrfd = 11
Telnet reqwill and reqdo's initialized
In modem_do_ring, ringing at 000000ed3fe70000
Telnet reqwill 1
Telnet reqdo 1
Telnet reqwill 3
Telnet reqdo 3
Send telnet reqs, rdwrfd=11

There is a comment at the top of scc.c

/* scc port 0 == channel A = slot 1 = c039/c03b */
/* port 1 == channel B = slot 2 = c038/c03a */

I am using slot 1, so I am not sure why c038 is being accessed when using
kegs 1.27:

Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!
Wr c038 to wr0 of 44!
Wr c038 to wr0 of 62, bad cmd cd:4!
Wr c038 to wr0 of 62!
Wr c038 to wr0 of 6a!

I don't get these "Wr c038" messages in kegs 1.23.

In article <AYKdnQq9X6fxfA35nZ2dnZfqn_GdnZ2d@earthlink.com>,
Hugh Hood <hughhood@earthlink.net> wrote:
>One more thing -- I was out of town this past weekend and was unable to
>do any more testing with my newly-installed KEGS on the Mac, but using a
>telnet window to connect to KEGS' port 6501, and then typing PR#1 in
>BASIC on the emulated Apple IIGS would indeed transfer any and all
>keystrokes typed on the emulated Apple IIGS to the telnet window, but it
>would not, for example, transfer the contents of the Catalog command
>typed from the same prompt.
>
>May I ask you to check that out on your machine?
>
>BTW, thanks for the quick response, and the damn-quick emulator.
>
>
>
>
>Hugh Hood

Yes, KEGS 1.27 had a new bug introduced with the "false read" change being
done incorrectly, and it messed up accessing the SCC registers. The ROM
would do "LDA $BFFD,Y", with Y=$3C, to try to access $C039. The "false read"
should go to $BF39 (which does nothing), but the KEGS bug had it read $C039.
The SCC clears state on reading this register, and so it broke the serial
port operation. The fix was to make the false read go to $BF39 as intended.

KEGS 1.28 at kegs.sourceforge.net fixes this bug, and adds faster ZipGS
speeds (up to 128MHz), along with other fixes.

Kent

On 6/21/2023 9:23 PM, Kent Dickey wrote:
>
> Yes, KEGS 1.27 had a new bug introduced with the "false read" change being
> done incorrectly, and it messed up accessing the SCC registers. The ROM
> would do "LDA $BFFD,Y", with Y=$3C, to try to access $C039. The "false read"
> should go to $BF39 (which does nothing), but the KEGS bug had it read $C039.
> The SCC clears state on reading this register, and so it broke the serial
> port operation. The fix was to make the false read go to $BF39 as intended.
>
> KEGS 1.28 at kegs.sourceforge.net fixes this bug, and adds faster ZipGS
> speeds (up to 128MHz), along with other fixes.
>
> Kent

Thank you, Kent. Can't wait to use it.

Hugh Hood

Thank you, thank you, thank you. kegs 1.28 works like a charm.
https://iili.io/HPcaRvp.png

On 6/22/2023 12:20 PM, mmphosis wrote:
> Thank you, thank you, thank you. kegs 1.28 works like a charm.
> https://iili.io/HPcaRvp.png
>

Agreed.

Yes, with KEGS v1.28 I'm now able to print from the emulated IIGS to a
network printer by running an ncat client-to-client relay between KEGS
Port 6501 and printer server Port 9100.

I have noticed one anomaly though, which I've tracked to a piece of code
in the scc_socket_driver.c file (which I don't fully understand the
workings of at this point).

After ncat makes its initial connection with Port 6501, KEGS sends out
the following 8-byte sequence, which of course the printer then prints
at the beginning of the first job.

That sequence is:

0xFB / 0x01 / 0xFD / 0x01 / 0xFB / 0x03 / 0xFD / 0x03

I'm pretty sure I can script the ncat connection script to discard this
before it relays it to the printer server, but I'm curious -- what is
the purpose of sending this sequence?

Thanks.

Hugh Hood

Here's what I've found out so far ...

It's telnet protocol.
https://datatracker.ietf.org/doc/html/rfc854

Telnet options.
https://www.iana.org/assignments/telnet-options/telnet-options.xhtml

1 Echo
https://www.rfc-editor.org/rfc/rfc857.html

3 Suppress Go Ahead
https://www.rfc-editor.org/rfc/rfc858.html

----- Depending on the telnet command(s), I believe those bytes are escaped
with 0xFF:
"Interpret as Command" (IAC) escape character 255

### convert to decimal
# for i in FB 01 FD 01 FB 03 FD 03; do printf "$((16#$i)) "; done | xargs
251 1 253 1 251 3 253 3

----- That sequence is:
$FB $01 $FD $01 $FB $03 $FD $03
251 1 253 1 251 3 253 3
will echo do echo will SGA do SUPPRESS-GO-AHEAD

It's some negotiation that we don't see when using the telnet program:

WILL ECHO

DO ECHO

WILL SUPPRESS-GO-AHEAD

DO SUPPRESS-GO-AHEAD

----- All this time, I've been using the telnet program

# man telnet | tail -n4
BUGS
The source code is not comprehensible.

Linux NetKit (0.17) August 15, 1999 Linux NetKit
(0.17)

----- I am also starting to mess around with netcat

### from the /dev/pts/1 terminal
# nc localhost 6501

### to adjust the terminal settings of /dev/pts/1 from another terminal
# sudo stty -F /dev/pts/1 -echo -ignbrk -brkint -ignpar -parmrk -inpck
-istrip -inlcr -igncr -icrnl -ixon -ixoff -icanon -opost -isig -iuclc
-ixany -imaxbel -xcase min 1 time 0

On Thursday, June 22, 2023 at 10:00:41 p.m. UTC-7, Hugh Hood wrote:
> On 6/22/2023 12:20 PM, mmphosis wrote:
> > Thank you, thank you, thank you. kegs 1.28 works like a charm.
> > https://iili.io/HPcaRvp.png
> >
> Agreed.
>
> Yes, with KEGS v1.28 I'm now able to print from the emulated IIGS to a
> network printer by running an ncat client-to-client relay between KEGS
> Port 6501 and printer server Port 9100.
>
> I have noticed one anomaly though, which I've tracked to a piece of code
> in the scc_socket_driver.c file (which I don't fully understand the
> workings of at this point).
>
> After ncat makes its initial connection with Port 6501, KEGS sends out
> the following 8-byte sequence, which of course the printer then prints
> at the beginning of the first job.
>
> That sequence is:
>
> 0xFB / 0x01 / 0xFD / 0x01 / 0xFB / 0x03 / 0xFD / 0x03
>
> I'm pretty sure I can script the ncat connection script to discard this
> before it relays it to the printer server, but I'm curious -- what is
> the purpose of sending this sequence?
>
> Thanks.
>
>
>
>
> Hugh Hood

Could it just be <SEND A CR> ? at the start.

# nc localhost 6501 > out.txt & sleep 1 && kill %1 && hexdump -vC out.txt &&
rm -f out.txt
[1] 65816
00000000 4e 4f 20 43 41 52 52 49 45 52 0d 0a 0d 0a 3f 53 |NO
CARRIER....?S|
00000010 59 4e 54 41 58 20 45 52 52 4f 52 07 0d 0a 5d 0a |YNTAX
ERROR...].|
00000020 ff fb 01 ff fd 01 ff fb 03 ff fd 03 |............|
0000002c
[1]+ Terminated nc localhost 6501 > out.txt

On Friday, June 23, 2023 at 2:39:38 p.m. UTC-7, mmphosis wrote:
> # nc localhost 6501 > out.txt & sleep 1 && kill %1 && hexdump -vC out.txt &&
> rm -f out.txt
> [1] 65816
> 00000000 4e 4f 20 43 41 52 52 49 45 52 0d 0a 0d 0a 3f 53 |NO
> CARRIER....?S|
> 00000010 59 4e 54 41 58 20 45 52 52 4f 52 07 0d 0a 5d 0a |YNTAX
> ERROR...].|
> 00000020 ff fb 01 ff fd 01 ff fb 03 ff fd 03 |............|
> 0000002c
> [1]+ Terminated nc localhost 6501 > out.txt
How about a CR at the End!?

On 6/23/2023 7:27 AM, mmphosis wrote:
> Here's what I've found out so far ...
>
> It's telnet protocol.
> https://datatracker.ietf.org/doc/html/rfc854
>

Agreed. You've hit the nail on it's head. Thanks.

FWIW - for those wondering what we're doing, here's the fast and
undeveloped looping script I've been using on the Windows command line
to funnel output from the GSPort variant of KEGS to my printer server on
my Windows machine:

**************

rem
rem Test Script for keeping ncat client-to-client relay
rem open;

echo off
:begin
ncat 127.0.0.1 6501 -c "ncat 192.168.1.170 9100"
goto begin

**************

When I've wanted to print from GSPort, I've opened the command line,
pasted this script, started GSPort, and then am able to 'print' from the
emulated IIgs (both text/PCL in ProDOS 8, and PostScript in GS/OS).

After I've quit GSPort, I just send a ctrl-C to the command line and exit.

It's not pretty or elegant, but hey it works.

I plan to do something much better on my Mac with KEGS (barring an
update from Kent enabling the feature I requested he consider) using a
bash (or AppleScript) script that would:

1. start KEGS;
2. start the ncat client-to-client relay (i.e. ncat 127.0.0.1 6501 -c
"ncat 192.168.1.170 9100");
and,
3. keep the relay looping until it detects that KEGS is no longer running.

Of course, if someone enterprising writes it for me before I 'get to
it', all the better. In the meantime, I'll just use the Mac Terminal to
enable printing from KEGS.

Hugh Hood

On 6/24/2023 1:45 PM, Hugh Hood wrote:
>
> When I've wanted to print from GSPort, I've opened the command line,
> pasted this script, started GSPort, and then am able to 'print' from the
> emulated IIgs (both text/PCL in ProDOS 8, and PostScript in GS/OS).
>

Let me correct myself. I don't paste that script. Rather, I've got it
saved as a .bat (batch) file and I just double-click it.

Hugh Hood

In article <hROdnat_nspyuwj5nZ2dnZfqnPWdnZ2d@earthlink.com>,
Hugh Hood <hughhood@earthlink.net> wrote:
>On 6/22/2023 12:20 PM, mmphosis wrote:
>> Thank you, thank you, thank you. kegs 1.28 works like a charm.
>> https://iili.io/HPcaRvp.png
>>
>
>Agreed.
>
>Yes, with KEGS v1.28 I'm now able to print from the emulated IIGS to a
>network printer by running an ncat client-to-client relay between KEGS
>Port 6501 and printer server Port 9100.
>
>I have noticed one anomaly though, which I've tracked to a piece of code
>in the scc_socket_driver.c file (which I don't fully understand the
>workings of at this point).
>
>After ncat makes its initial connection with Port 6501, KEGS sends out
>the following 8-byte sequence, which of course the printer then prints
>at the beginning of the first job.
>
>That sequence is:
>
>0xFB / 0x01 / 0xFD / 0x01 / 0xFB / 0x03 / 0xFD / 0x03
>
>I'm pretty sure I can script the ncat connection script to discard this
>before it relays it to the printer server, but I'm curious -- what is
>the purpose of sending this sequence?

Since the 90's, KEGS has supported you doing "telnet localhost 6501" to
connect to the slot 1 serial port. Telnet is not a "transparent" protocol,
KEGS HAS to do some things to get character echo turned off and immediate
sending of characters to work. So it sends these Telnet commands.

Everyone knew telnet was insecure, but then someone seemed to think they
just discovered a new problem with telnet, and got it defeatured. And
then it seems netcat was born (or found, I never heard of it before about
10 years ago though).

And netcat has a problem--it doesn't know about the Telnet protocol and
so just ignores it. So then it passes these weird escape codes on to
whatever you connect netcat to.

So, I have 3 options in KEGS:

1) Do nothing. Netcat will work poorly (it will see these Telnet escape
codes at startup) and telnet (if you can find it) will work like
it always did.
2) Remove the Telnet codes. Netcat will work as expected, but now Telnet
will not really work--you'll see extra chars and what you type
will only get sent to KEGS when you press return.
3) Someone give me some ideas on how to detect Telnet vs. Netcat connecting
to KEGS.

Kent

On 6/25/2023 9:49 PM, Kent Dickey wrote:
>
> So, I have 3 options in KEGS:
>
> 1) Do nothing. Netcat will work poorly (it will see these Telnet escape
> codes at startup) and telnet (if you can find it) will work like
> it always did.
> 2) Remove the Telnet codes. Netcat will work as expected, but now Telnet
> will not really work--you'll see extra chars and what you type
> will only get sent to KEGS when you press return.
> 3) Someone give me some ideas on how to detect Telnet vs. Netcat connecting
> to KEGS.
>

Kent,

I suggest you do nothing. Besides, ncat (a successor to NetCat) has a
Telnet option as follows:

-t, --telnet (Answer Telnet negotiations)
Handle DO/DONT WILL/WONT Telnet negotiations. This makes it possible
to script Telnet sessions with Ncat.

My only concern would be that unless Telnet is put in binary mode (and
I'm not sure if the KEGS Telnet server supports that) it would not be
adequate to say send 8-bit graphic data to a printer over ncat. That's
all. {And, I don't foresee a need to do that}.

I do send 8-bit characters to my network printer via the same method,
but they are in the printable 191 characters of the ISO 8859 Latin code
page, and as I understand it, Telnet must allow transmission of those
191 characters even when not in binary mode.

Tomorrow I'll try ncat with the -t option and see if it suppresses the
forwarding of the Telnet negotiation codes sent by KEGS.

Even if it doesn't, I'm pretty sure I can script to discard them.

Thanks for looking into this. Again, I wouldn't change anything.

Hugh Hood