Paul wrote:

https://en.wikipedia.org/wiki/Cygwin

That turned out to be harder than it needed to be.

The idea was, you use the "setup.exe" to populate
a C:\cygwin directory with stuff.

It looks like a year ago, they decided to drop support
for WinXP. If you want Cygwin on a WinXP machine now,
the recipe is here.

http://www.crouchingtigerhiddenfruit...wintimemachine

The setup.exe files for the WinXP version are here.

https://web.archive.org/web/20160728...://cygwin.com/

"Install it by running setup-x86.exe (32-bit installation)
or setup-x86_64.exe (64-bit installation)."

When you download that, you do this in a Command Prompt.

setup-x86.exe -X

and on any future invocation of the setup (like if you need
to download another package of software later), you do this again
in a Command Prompt window

setup-x86.exe -X

The Time Machine page gives a URL that goes into the
mirror site dialog, pointing the installer at the
(private) web server with repository on it.

*******

If you wanted to install Cygwin for an OS later than
WinXP, you follow the instructions on the Cygwin site.

http://www.cygwin.com/

I got the WinXP ddrescue package installed. I made a couple
fake disk drives in a virtual machine and tested it.
This is my fake environment

disk 1 WinXP SP3 x32 /dev/sda
disk 2 blank 1GB disk drive /dev/sdb
disk 3 blank 1GB disk drive /dev/sdc

And in Disk Management, they're listed in that order too.

This is the content of the logfile (aka mapfile) after
a successful (single) run. Since my fake disks don't
have CRC errors, it's not possible to test that the program
works with damaged disks.

This is my sdb.log file, after the run. If the source
disk (sdb) had CRC errors, then additional lines in this
file would say "it hasn't copied them yet". And a second
or subsequent pass would be required.

# Mapfile. Created by GNU ddrescue version 1.21
#
# Command line: ddrescue -b8M --force /dev/sdb /dev/sdc sdb.log
#
# Start time: 2017-10-11 00:31:52
# Current time: 2017-10-11 00:32:17
# Finished
# current_pos current_status
0x3FE56C00 +
# pos size status
0x00000000 0x3FFC0000 +

The actual "device" size is 1,073,479,680 bytes.
In hex that is 0x3FFC0000. So it looks like the last
line in the log file, shows the job as completed.

I don't know what the "current_pos" means. It's
2890 sectors from the end.

Paul