GSPort, Uthernet and TreeHugger?

[Hugh Hood]

I finally got around to using the pre-configured disk image (thanks Ewen
and David) for GSPort that contains Marinetti and the Uthernet LL.

I installed Win10Pcap on my Windows 10 box and configured the TCP/IP
Control Panel.

Apparently, I did this correctly because I've been using Ewen's SAFE2 to
FTP both to various internet sites as well as my own networked LaserJet
print server. So far, so good.

But, I cannot get Daniel Krue's TreeHugger to communicate with the same
LaserJet networked printer despite my best efforts.

I tried it with both my Direct Connect PostScript driver (patched
LaserWriter driver) as well as with the Harmonie LJIII driver. I get
absolutely nothing. No activity at all. But, the GS does not lock and it
would appear that the job was 'sent'.

I know that Chris report good results with TreeHugger with a real GS and
Uthernet card, so I'm at a loss as to what might be wrong.

Has anyone here used TreeHugger with GSPort and its emulated Uthernet
feature?

If not, and you're able, may I ask you to give it a try and post back?

Thanks. I haven't tried Krue's earlier versions of TreeHugger, but I
might just in case.




Hugh Hood

[Hugh Hood]

Can any of you network savvy guys offer some insight into why the
emulated Uthernet in GSPort is having issues when using the TreeHugger
TCP/IP printer port driver?

Using Kelvin Sherlock's TCP Snooper CDA, it shows that an ipid has been
assigned, that the connection has been 'Established' to the proper
addresses and ports, and that the Send Queue has content, but nothing
'happens'.

So, I did a Wireshark dump on the activity between GSPort
(192.168.1.3:1025) and the network printer (192.168.1.170:9100), and it
appears that data is actually being sent, but for some reason unknown to
me, it is being re-transmitted as though it's not being acknowledged.

Here's part of the dump:

Source Destination Proto Len Info
--------------------------------------------------------------------------
192.168.1.3 192.168.1.170 TCP 54 1025 → 9100 [SYN] Seq=0 Win=16384 Le
192.168.1.3 192.168.1.170 TCP 54 1025 → 9100 [ACK] Seq=1 Ack=1 Win=16
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 TCP 1454 [TCP Window Full] 1025 → 9100 [ACK]
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 TCP 1454 [TCP Window Full] [TCP Retransmissio


What surprises me is that I assumed I would see some communication
backward from the printer to GSPort, but it all appears to be one-way.
Contrast that to printing from Windows applications where clearly
Wireshark shows comm in both directions.

And, just to make certain that this isn't a firewall problem, I disabled
it completely for testing.

At this point, I'm way beyond my network knowledge skills. Any suggestions?

Thanks.





Hugh Hood

Post by Hugh Hood
But, I cannot get Daniel Krue's TreeHugger to communicate with the same
LaserJet networked printer despite my best efforts.
I tried it with both my Direct Connect PostScript driver (patched
LaserWriter driver) as well as with the Harmonie LJIII driver. I get
absolutely nothing. No activity at all. But, the GS does not lock and it
would appear that the job was 'sent'.

[Andrew Roughan]

Post by Hugh Hood
What surprises me is that I assumed I would see some communication
backward from the printer to GSPort, but it all appears to be one-way.
Contrast that to printing from Windows applications where clearly
Wireshark shows comm in both directions.

Hi Hugh,

My guess is that the printer is dropping the packets for some reason.
Perhaps it needs a certain option that is not implemented by Marinetti.
Can you capture an exchange that does work and then we can see what it is
doing that may be different?

Regards
Andrew

[Hugh Hood]

Thank you Andrew,

I know that you know Marinetti, so here goes the Wireshark dump from the
beginning of an exchange between a Windows application (Thunderbird) and
the same Port 9100 (RAW) printer:

Source Destination Proto Len Info
--------------------------------------------------------------------------
192.168.1.2 192.168.1.170 TCP 66 62072 → 9100 [SYN] Seq=0 Win=64240
192.168.1.170 192.168.1.2 TCP 60 9100 → 62072 [SYN, ACK] Seq=0 Ack=1
192.168.1.2 192.168.1.170 TCP 54 62072 → 9100 [ACK] Seq=1 Ack=1

192.168.1.2 192.168.1.170 TCP 5894 [TCP Window Full] 62072 → 9100 [ACK
192.168.1.170 192.168.1.2 TCP 60 9100 → 62072 [ACK] Seq=1 Ack=5841 W
192.168.1.2 192.168.1.170 TCP 2974 [TCP Window Full] 62072 → 9100 [ACK
192.168.1.170 192.168.1.2 TCP 60 9100 → 62072 [ACK] Seq=1 Ack=7301 W
192.168.1.2 192.168.1.170 TCP 2974 [TCP Window Full] 62072 → 9100 [ACK
192.168.1.170 192.168.1.2 TCP 60 9100 → 62072 [ACK] Seq=1 Ack=11681
{continues}

Unlike what is happening with TreeHugger, Marinetti and GSPort, we have
the back and forth SYN/ACK for the first (3) items, and then a series of
print data sends and acknowledgments until the job completes.

Is there a chance that there is no issue at all with either TreeHugger
and/or Marinetti, but rather it is due to the WinPcap packet capture
program that GSPort relies upon?

What puzzles me is that FTPing to the same printer (different port) with
Ewen's SAFE2 within GSPort works just fine.

Thanks for looking into this. Maybe getting Krue's TreeHugger to work
from GSPort is a lost cause for the time being, but before I moved on
with my testing on the Direct Connect PostScript driver I wanted to give
it my best shot. I don't have an Uthernet card or I would just test the
driver out on a real IIgs with TreeHugger.




Hugh Hood

Post by Andrew Roughan

Post by Hugh Hood
What surprises me is that I assumed I would see some communication
backward from the printer to GSPort, but it all appears to be one-way.
Contrast that to printing from Windows applications where clearly
Wireshark shows comm in both directions.

Hi Hugh,
My guess is that the printer is dropping the packets for some reason.
Perhaps it needs a certain option that is not implemented by Marinetti.
Can you capture an exchange that does work and then we can see what it is
doing that may be different?
Regards
Andrew

[Hugh Hood]

Post by Hugh Hood
What puzzles me is that FTPing to the same printer (different port) with
Ewen's SAFE2 within GSPort works just fine.

OK, I captured Ewen's SAFE2 establishing comm with the printer and
something interesting appeared. Lines 2 & 3 show the ARP (address
resolution protocol) request to supply the mac address of GSPort back to
the printer.

Source Destination Proto Len Info
-----------------------------------------------------------------------
192.168.1.3 192.168.1.170 TCP 54 1026 → 21 [SYN] Seq=0 Win=16384 Len
HewlettP_bf:f8:60 Broadcast ARP 60
Who has 192.168.1.3? Tell 192.168.1.170
CirrusLo_a2:a2:a2 HPbf:f8:60 ARP 42
192.168.1.3 is at 00:0e:3a:a2:a2:a2
192.168.1.170 192.168.1.3 TCP 60 21 → 1026 [SYN, ACK] Seq=0 Ack=1 Wi
192.168.1.3 192.168.1.170 TCP 54 1026 → 21 [ACK] Seq=1 Ack=1 Win=163

Is this ARP exchange occurring due to something in Ewen's coding, or is
this something Marinetti handles?

{Ed - it certainly seems we would be seeing 2-way comm. No wonder it's
failing.}




Hugh Hood

[eeastman]

Post by Hugh Hood
Can any of you network savvy guys offer some insight into why the
emulated Uthernet in GSPort is having issues when using the TreeHugger
TCP/IP printer port driver?
Using Kelvin Sherlock's TCP Snooper CDA, it shows that an ipid has been
assigned, that the connection has been 'Established' to the proper
addresses and ports, and that the Send Queue has content, but nothing
'happens'.
So, I did a Wireshark dump on the activity between GSPort
(192.168.1.3:1025) and the network printer (192.168.1.170:9100), and it
appears that data is actually being sent, but for some reason unknown to
me, it is being re-transmitted as though it's not being acknowledged.
Source Destination Proto Len Info
--------------------------------------------------------------------------
192.168.1.3 192.168.1.170 TCP 54 1025 → 9100 [SYN] Seq=0 Win=16384 Le
192.168.1.3 192.168.1.170 TCP 54 1025 → 9100 [ACK] Seq=1 Ack=1 Win=16
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 TCP 1454 [TCP Window Full] 1025 → 9100 [ACK]
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 IPv4 1514 Fragmented IP protocol (proto=TCP 6,
192.168.1.3 192.168.1.170 TCP 1454 [TCP Window Full] [TCP Retransmissio
What surprises me is that I assumed I would see some communication
backward from the printer to GSPort, but it all appears to be one-way.
Contrast that to printing from Windows applications where clearly
Wireshark shows comm in both directions.
And, just to make certain that this isn't a firewall problem, I disabled
it completely for testing.
At this point, I'm way beyond my network knowledge skills. Any suggestions?
Thanks.
Hugh Hood

Post by Hugh Hood
But, I cannot get Daniel Krue's TreeHugger to communicate with the same
LaserJet networked printer despite my best efforts.
I tried it with both my Direct Connect PostScript driver (patched
LaserWriter driver) as well as with the Harmonie LJIII driver. I get
absolutely nothing. No activity at all. But, the GS does not lock and it
would appear that the job was 'sent'.

So shouldn’t it be communicating on port 9100 both ways?

[Christopher G. Mason]

Post by Hugh Hood
I finally got around to using the pre-configured disk image (thanks Ewen
and David) for GSPort that contains Marinetti and the Uthernet LL.
I installed Win10Pcap on my Windows 10 box and configured the TCP/IP
Control Panel.

Win10Pcap and the recommended replacement for it, npcap don't work
correctly with GSport. Spin up a Windows 7 VM, install WinPcap 4.1.3 and
test there. TreeHugger does work with GSport as I have tested it in the
past.

[Hugh Hood]

Chris,

Thanks for the input.

Concerning Win10Pcap -- Yes, I noticed right off the bat that something
wasn't quite right when the GSPort console was issuing error messages
during TCP activities. Surprisingly, though, Ewen's SAFE2 worked fine in
spite of that.

So, I uninstalled Win10Pcap and installed WinPcap in Windows 10, based
on reports that it still worked fine in Windows 10.

GSPort does not report errors when using it, and SAFE2 continues to
work, so I'll stick with it. And, while I appreciate your suggestion to
set up a Windows 7 virtual machine, my enthusiasm for doing that just
for this project testing is pretty low, I'm afraid.

Regarding TreeHugger -- I _still_ haven't gotten that to work, and that
really bugs me. I've done Wireshark dumps and fiddled with MTU settings
in GSPort, the host Windows 10 computer, and on my router, and I've
still nothing to show for it.

Wireshark (if it can be believed) is showing that the packet size coming
from TreeHugger is 6,000 bytes, which seems like it would be 4X what it
should be. I guess I'll keep fiddling, not because it's worth it, but
because it bugs me when things don't work.

BTW, are you able to test TreeHugger with GSPort currently and let me
know how it goes? Thanks.




Hugh Hood

Post by Christopher G. Mason

Post by Hugh Hood
I finally got around to using the pre-configured disk image (thanks Ewen
and David) for GSPort that contains Marinetti and the Uthernet LL.
I installed Win10Pcap on my Windows 10 box and configured the TCP/IP
Control Panel.

Win10Pcap and the recommended replacement for it, npcap don't work
correctly with GSport. Spin up a Windows 7 VM, install WinPcap 4.1.3 and
test there. TreeHugger does work with GSport as I have tested it in the
past.

[Christopher G. Mason]

Post by Hugh Hood
BTW, are you able to test TreeHugger with GSPort currently and let me
know how it goes? Thanks.

I did exactly that today while packet tracing the derpy LaserWriter
driver's AppleTalk comms. I was able to get TreeHugger printing to my HP
color LaserJet. No drama whatsoever.

[Hugh Hood]

Very interesting. Thanks for going to that trouble for me.

A few more questions, if I may ...

Was this on Windows 10?

Was this GSPort v0.31?

I assume you're using one of the Harmonie drivers?

Is this version 1.0 of TreeHugger?

Is there a wired router between the LaserJet and the machine running GSPort?




Hugh Hood

P.S. - I don't know whether it would help you with your AppleTalk /
Netatalk / LaserWriter issue, but the Direct PostScript driver I've been
testing and using dispenses with the back-and-forth exchanges and also
adds a few things like some updated PostScript code, sending a
<Control-D> at the end of the job and ensuring that the _first_ line of
the job begins with '%!PS-Adobe-2.0'. If you like to try it for grins
with the AppleTalk port driver, let me know.

Post by Christopher G. Mason

Post by Hugh Hood
BTW, are you able to test TreeHugger with GSPort currently and let me
know how it goes? Thanks.

I did exactly that today while packet tracing the derpy LaserWriter
driver's AppleTalk comms. I was able to get TreeHugger printing to my HP
color LaserJet. No drama whatsoever.

[Christopher G. Mason]

Post by Hugh Hood
Very interesting. Thanks for going to that trouble for me.
A few more questions, if I may ...
Was this on Windows 10?
Was this GSPort v0.31?
I assume you're using one of the Harmonie drivers?
Is this version 1.0 of TreeHugger?
Is there a wired router between the LaserJet and the machine running GSPort?

GSport v0.31 running in Windows 7. Everything wired via Ethernet. Note
that WinPCap doesn't work over WiFi adapters.

[Hugh Hood]

Post by Hugh Hood
Wireshark (if it can be believed) is showing that the packet size coming
from TreeHugger is 6,000 bytes, which seems like it would be 4X what it
should be. I guess I'll keep fiddling, not because it's worth it, but
because it bugs me when things don't work.

OK,

I finally got past TreeHugger/Marinetti on GSPort (Windows 10) sending
6,000-byte packets (that had to be fragmented) to my LaserJet/JetDirect
print server.

It may not be kosher, but for testing purposes I patched four bytes in
Ewen's Uthernet (original) link layer and now I have an MSS=1400 which
this older (circa 2003) print server apparently can handle.

But, now I'm getting duplicate ACK's being sent from GSPort to the print
server, and after (2) of those, Wireshark shows no further communication
between GSPort and the print server.

For reference, GSPort is 192.168.1.180:1026, while the
LaserJet/JetDirect print server is 192.168.1.170:9100.

--------------------------------------------------------------------
192.168.1.180 192.168.1.170 TCP 58 1026 > 9100 [SYN] Seq=0 Win=16384
Len=0 MSS=1400

192.168.1.170 192.168.1.180 TCP 60 9100 > 1026 [SYN, ACK] Seq=0 Ack=1
Win=0 Len=0 MSS=1400

192.168.1.180 192.168.1.170 TCP 58 1026 > 9100 [ACK] Seq=1 Ack=1
Win=16384 Len=0 MSS=1400

192.168.1.180 192.168.1.170 TCP 58 [TCP Dup ACK 4#1] 1026 > 9100 [ACK]
Seq=1 Ack=1 Win=16384 Len=0 MSS=1400

192.168.1.180 192.168.1.170 TCP 58 [TCP Dup ACK 4#2] 1026 > 9100 [ACK]
Seq=1 Ack=1 Win=16384 Len=0 MSS=1400
--------------------------------------------------------------------

Should I be looking for a network timing problem? The Windows 10 machine
is set to auto-negotiate, but I know the LaserJet/JetDirect is 10/100.
Everything works fine when printing from Windows apps however.




Hugh Hood

[Hugh Hood]

Good news for those here who are tired of seeing this seemingly unending
thread -- Let's put it to bed!

I have finally resolved the issue I've been having trying to use Daniel
Krue's TreeHugger TCP/IP Printer Port Driver from within GSPort running
under Windows 10.

I'm happy to report that not only does Krue's TreeHugger work with the
LaserJet drivers from the Harmonie package in GSPort/Windows 10/WinPcap,
it also works with the Direct Connect PostScript (patched LaserWriter)
printer driver that I've been testing.

Thanks so much for the input I've received on this, both on the group
and off.

Although I suspect this issue may only affect an audience of one (me),
here's what I did to 'fix' the problem. It involves reducing the segment
size sent by Marinetti, and is done by patching (4) bytes within Ewen
Wannop's Uthernet (original) Link Layer.

{Disclaimer - Ewen's software is actively supported, is bug-free and I
consider it somewhat irreverent to even mention patching it, but I did
this in order to overcome a very strange issue with a 2003-vintage
600N/3113A JetDirect print server in an HP laser printer, and should
that issue ever affect someone else, I'll leave this here in the hope
that it will be helpful. My methodology was trial, error and hunch, and
was NOT done out of a thorough understanding of either Marinetti nor the
Uthernet Link Layer.}

OK. Using ChewBagger (of course), open the data fork of the UTHERNET
file, and change the following bytes:

BYTE FROM TO COMMENT
----------------------------------------------
$1229 $DC $40 Change 1500 to 320
$122A $05 $01

$1352 $DC $40 Change 1500 to 320
$1353 $05 $01

Save, quit ChewBagger, GS/OS and GSPort. Restart and you should be good.

Although I reduced the size from 1500 down to 320, I suspect it can be
slightly higher than that and still work, although I know it failed when
set to 576.





Hugh Hood

[Antoine Vignau]

That is good news, Hugh. Congrats for finding a way to solve your problem but...

...could this happen to others? Is reducing the segment sent by Marinetti the real root cause? Should the size be set manually/automagically? Why does it work on all machines but ours?

Thank you,
Antoine

[Hugh Hood]

Post by Antoine Vignau
...could this happen to others? Is reducing the segment sent by
Marinetti the real root cause? Should the size be set
manually/automagically? Why does it work on all machines but ours?

Antoine,

Those are great questions to which I really don't have the answers.

Here's the 'chain':

TreeHugger -> Uthernet LL -> Marinetti -> GSPort Emulated Uthernet ->
WinPcap -> NIC on host Windows 10 machine -> router -> print server.

That's quite a few components that are working together, and since my
networking knowledge has been newly and hurriedly learned, I would defer
to guys like Wannop, Roughan, Schmidt, Schmidt, Finnigan and Sherlock
for intelligent answers to your questions.

In the meantime, I'll just have to be content with knowing that my hack
works for TreeHugger and my print server, rather than knowing why it
works. I did partially disassemble TreeHugger and made a few test
patches hoping that they would 'fix' the issue, but only the segment
size patches in the Uthernet LL proved to be effective in my case.




Hugh Hood

[Andrew Roughan]

I’ve asked Ewen to take a look at exposing a configuration setting. I
thought there was one already.

Regards
Andrew

[Hugh Hood]

Thank you, Andrew.

Post by Andrew Roughan
I’ve asked Ewen to take a look at exposing a configuration setting. I
thought there was one already.
Regards
Andrew

[Andrew Roughan]

Post by Andrew Roughan
I’ve asked Ewen to take a look at exposing a configuration setting. I
thought there was one already.

https://sourceforge.net/p/marinetti/feature-requests/40/

[Hugh Hood]

Ewen has graciously and swiftly released an updated Marinetti link layer
for the original Uthernet card.

This version, 1.0.4, now allows setting the MTU in the configuration
dialog, just as the configuration dialogs in the Uthernet 2 and the
Sweet16 link layers allow.

It is available from Ewen's site:

<http://speccie.uk/speccie/downloads/uthernetll.bxy>

Before configuring the new link layer, be sure to delete the
'Preferences' and 'link layer data' files from the System/TCP directory.

To recap and relate this back to this particular thread -- I've found it
necessary to set the Uthernet link layer MTU to '320' in order to use
Daniel Krue's 'TreeHugger' GS/OS Printer Port driver (AppSocket /
JetDirect / Raw-Port9100 protocol) when running GSPort under Windows 10
using its emulated Uthernet card and printing to a networked HP LaserJet
JetDirect 600N/3113A print server.

TreeHugger is available from Krue's site:

<http://krue.net/treehugger/>


Thank you Ewen, and thank you Andrew.





Hugh Hood

Post by Andrew Roughan

Post by Andrew Roughan
I’ve asked Ewen to take a look at exposing a configuration setting. I
thought there was one already.

https://sourceforge.net/p/marinetti/feature-requests/40/

[Antoine Vignau]

and thank you the three of you!
av

[Ewen]

Antoine,

Post by Antoine Vignau
and thank you the three of you!

I feel fully thanked... :-)

Cheers - Ewen

[Christopher G. Mason]

Post by Hugh Hood
To recap and relate this back to this particular thread -- I've found it
necessary to set the Uthernet link layer MTU to '320' in order to use
Daniel Krue's 'TreeHugger' GS/OS Printer Port driver (AppSocket /
JetDirect / Raw-Port9100 protocol) when running GSPort under Windows 10
using its emulated Uthernet card and printing to a networked HP LaserJet
JetDirect 600N/3113A print server.

FWIW, I am running a JetDirect 635N along with Windows 7 and GSport 0.31
with the last revision of WinPCap and AppSocket printing works fine with
the default MTU. Never had a problem with it. It also worked with the
JetDirect 615N card until the card failed like the rest of them did
(known hardware flaw). Printing via the LaserWriter driver using the
AppleTalk bridge works as well. A CUPS server with port 9100 open and
GhostPCL setup as an input filter should work as well.... it might even
work with Harmonie's PCL color DeskJet drivers.

[Hugh Hood]

Post by Christopher G. Mason
FWIW, I am running a JetDirect 635N along with Windows 7 and GSport 0.31
with the last revision of WinPCap and AppSocket printing works fine with
the default MTU. Never had a problem with it. It also worked with the
JetDirect 615N card until the card failed like the rest of them did
(known hardware flaw).

Yes. To satisfy my curiosity, I'd really like to find someone who is
using GSPort under Windows 10 (as opposed to 7) along with the JetDirect
600N in order to compare results and determine if their experience is
different from mine.

Whether I ever will find that person remains to be seen. BTW, (I 'new'
to Windows), is Windows 7 still being supported by Microsoft? I ask
because I still run across quite a few people who are using it.




Hugh Hood

[Christopher G. Mason]

Post by Hugh Hood
Whether I ever will find that person remains to be seen. BTW, (I 'new'
to Windows), is Windows 7 still being supported by Microsoft? I ask
because I still run across quite a few people who are using it.

Windows 7 reached EOL in January. GSport needs to be updated for NPCap
eventually. GSplus has those changes, but the pre-compiled binaries lack
networking support!

[David Schmidt]

Post by Christopher G. Mason

Post by Hugh Hood
Whether I ever will find that person remains to be seen. BTW, (I 'new'
to Windows), is Windows 7 still being supported by Microsoft? I ask
because I still run across quite a few people who are using it.

Windows 7 reached EOL in January. GSport needs to be updated for NPCap
eventually. GSplus has those changes, but the pre-compiled binaries lack
networking support!

I've been meaning to devote some cycles to an updated build of GSport.
Maybe NPCap is a good excuse to get moving. Is there a PR for GSplus
that has NPCap stuff? I'm not sure where those changes are.

[Kelvin Sherlock]

https://nmap.org/npcap/guide/npcap-devguide.html#npcap-feature-native-dll

Basically call SetDllDirectory("C:\\Windows\\System32\Npcap\\") so it will
use the new wpcap.dll

Post by David Schmidt

Post by Christopher G. Mason

Post by Hugh Hood
Whether I ever will find that person remains to be seen. BTW, (I 'new'
to Windows), is Windows 7 still being supported by Microsoft? I ask
because I still run across quite a few people who are using it.

Windows 7 reached EOL in January. GSport needs to be updated for NPCap
eventually. GSplus has those changes, but the pre-compiled binaries lack
networking support!

I've been meaning to devote some cycles to an updated build of GSport.
Maybe NPCap is a good excuse to get moving. Is there a PR for GSplus
that has NPCap stuff? I'm not sure where those changes are.

-------
ProLine: ***@pro-kegs

[Christopher G. Mason]

Post by Kelvin Sherlock
https://nmap.org/npcap/guide/npcap-devguide.html#npcap-feature-native-dll
Basically call SetDllDirectory("C:\\Windows\\System32\Npcap\\") so it will
use the new wpcap.dll

For such minor changes, I'm surprised GSport doesn't work with NPcap in
WinPcap compatibility mode on Windows 10. I had trouble with using other
older WinPcap software with NPcap as well (Basilisk II fork with WinPcap
networking).